Skip to content

🙌 OpenHands: AI-Driven Development

docs(verify-openhands): map maintenance 2026-10-19 (#18658)

Engel Nyst8fa496c1dad4
diff --git a/.agents/skills/verify-openhands/references/feature-map/F02-app-shell.md b/.agents/skills/verify-openhands/references/feature-map/F02-app-shell.md
index e329672b36..a4c9170ffb 100644
--- a/.agents/skills/verify-openhands/references/feature-map/F02-app-shell.md
+++ b/.agents/skills/verify-openhands/references/feature-map/F02-app-shell.md
@@ -26,7 +26,7 @@ Source: `src/components/features/sidebar/`, `src/components/features/command-men
- `F02.command-menu-search`: every search term must match the title, description or keywords. **No commands found** is the empty state. The X button clears the query. Arrow keys wrap around the list, hovering an item selects it, Enter runs the selected item, and Enter does nothing when nothing matches.
- `F02.command-menu-new-tab`: a middle-click or modifier-click on a **Go** item opens the route in a new tab and leaves the menu open.
- `F02.command-menu-phone`: at 390 px the menu fits the screen and opens from the drawer's Search row. Escape closes the menu and leaves the drawer open. Running an item closes the drawer.
-- `F02.mobile-drawer`: at phone width the rail is hidden and a hamburger (`sidebar-mobile-menu-toggle`) opens a 300 px drawer. The hamburger is in the top bar, or in the chat header on conversation pages. The drawer closes on its chevron, Escape, a backdrop click or any route change.
+- `F02.mobile-drawer`: at phone width the rail is hidden and a hamburger (`sidebar-mobile-menu-toggle`) opens a 300 px drawer. The hamburger is in the top bar, or in the chat header on conversation pages. The drawer closes on its chevron, Escape, a backdrop click or any route change. A modal opened over the drawer takes Escape first: Escape closes only that modal (or nothing, for one that ignores Escape) and the drawer stays open.
- `F02.mobile-top-bar`: at phone width, settings sub-pages and Customize detail pages (`/skills`, `/mcp`, `/plugins`, `/apps`) show a Back chevron that returns to `/settings` or `/customize`.
- `F02.checklist`: after onboarding, the expanded rail shows a **Getting started** card with **N complete** and six items. They link to `/settings/llm`, `/conversations`, `/automations`, `/settings/agents`, `/mcp` and the external Slack invite.
- `F02.checklist-item-links`: clicking an item's row navigates in the same tab to its route (for example **Schedule a task** opens `/automations`) without marking the item done.
@@ -41,7 +41,7 @@ Source: `src/components/features/sidebar/`, `src/components/features/command-men
- `F02.update-tile`: when npm `latest` is newer than the running version, a **New version** tile (`agent-canvas-version-tile`) opens an update modal with the npm and Docker commands. The tile is never shown in locked-Cloud mode.
- `F02.alert-banner`: when the server config reports maintenance, faulty models or an error message, a dismissible banner (`alert-banner`) appears at the top of the content area.
- `F02.settings-404-modal`: if the backend's settings return 404, the LLM settings modal opens by itself. Other settings fetch errors show a toast.
-- `F02.error-toasts`: failed requests show a top-right error toast with the product or server message, and the toast closes by itself. Toasts stay readable in every colour theme.
+- `F02.error-toasts`: failed requests show a top-right error toast with the product or server message and a circled-X icon in the theme's status error colour, and the toast closes by itself. Toasts stay readable in every colour theme.
## How to get to it (user POV)
@@ -75,7 +75,7 @@ Preconditions:
- **Search, empty state, keyboard (`F02.command-menu-search`).** The Toggle sidebar action closed the menu: open it with `control-openhands browser press Control+k`, then run `control-openhands browser fill 'testid=command-menu >> role=combobox' theme` and `control-openhands browser count 'testid=command-menu >> role=option'` (`1`, Application, matched by its keywords). The term `settings secret` leaves `1` option, `secret automations` leaves `0` (all terms must match) and `model` leaves `3` (MCP servers, LLM profiles, Model Router). Fill `qa-zzz-nothing`, then run `control-openhands browser text 'testid=command-menu >> role=listbox'` (`No commands found` / `Try a page, action, or setting name.`) and `control-openhands browser screenshot --feature F02.command-menu-search --name no-results`. Run `control-openhands browser press Enter`; the menu stays open and the URL does not change. Run `control-openhands browser click 'testid=command-menu >> role=button[name="Clear search"]'`; the value is `""` and the count is back to `14`. Keyboard: `control-openhands browser attr 'testid=command-menu >> role=combobox' aria-activedescendant` starts at `command-menu-option-new-chat`. `control-openhands browser press ArrowUp` wraps to `command-menu-option-toggle-sidebar`, and `ArrowDown` returns to `new-chat`, then moves to `customize`. `control-openhands browser hover 'testid=command-menu >> role=option[name=/Secrets settings/]'` sets the value to `command-menu-option-secrets-settings`.
- **Open in a new tab (`F02.command-menu-new-tab`).** With the menu open and `secrets` typed, `control-openhands browser attr 'testid=command-menu >> role=option[name=/Secrets settings/]' href` is `/settings/secrets`. Run `control-openhands browser click 'testid=command-menu >> role=option[name=/Secrets settings/]' --button middle` and `control-openhands browser tabs`; tab 1 is `/settings/secrets`, tab 0 keeps its URL, and `browser count 'testid=command-menu'` is still `1`. Run `control-openhands browser close-tab 1`. The modifier path works the same: `control-openhands browser click 'testid=command-menu >> role=option[name=/Secrets settings/]' --modifiers Control` opens tab 1 with the menu still open; close that tab too. The click leaves focus on the option; `control-openhands browser press Escape` still closes the menu (count `0`).
- **Keyboard focus (`F02.command-menu-focus`).** From `/conversations` run `control-openhands browser click 'testid=chat-input'`, `control-openhands browser press Control+k`, `browser count 'testid=command-menu'` (`1`) and `control-openhands browser eval "document.activeElement && document.activeElement.id"` (`command-menu-search`): the shortcut opens the menu from inside the composer. Run `control-openhands browser press Tab`; the same `eval` returns `command-menu-option-new-chat`. `control-openhands browser press Escape` closes the menu from the option (count `0`), and `control-openhands browser eval "document.activeElement.dataset.testid"` returns `chat-input`: focus is back in the composer.
-- **Phone drawer (`F02.mobile-drawer`).** Run `control-openhands browser viewport phone` and `control-openhands browser goto /conversations`. `control-openhands browser visible 'aside[data-collapsed]'` is `false`. Run `control-openhands browser click 'testid=sidebar-mobile-menu-toggle'` and `control-openhands browser wait 'testid=sidebar-mobile-drawer'`. `control-openhands browser bbox 'testid=sidebar-mobile-drawer'` gives width `300`, `insideViewport` `true` and `pageHorizontalOverflow` `false`, and `control-openhands browser screenshot --feature F02.mobile-drawer --name open` shows the full rail. Close paths, each followed by `control-openhands browser wait 'testid=sidebar-mobile-drawer' --state detached`: `control-openhands browser click 'testid=sidebar-mobile-drawer-close'`; `control-openhands browser press Escape`; `control-openhands browser click 'testid=sidebar-mobile-drawer >> testid=sidebar-automations-link'` (route change; the URL becomes `/automations`); and a backdrop click to the right of the drawer, `control-openhands browser mouse-click 350 400` (on `/conversations`, `control-openhands browser click 'testid=submit-button' --force` also lands on the backdrop, because the empty composer's send button sits at x≈326 under it). On a conversation page (`browser goto` the URL from `conversation start`), `control-openhands browser bbox 'testid=sidebar-mobile-menu-toggle'` is at the top-left of the chat header. `control-openhands browser screenshot --feature F02.mobile-drawer --name conversation-header` shows it beside the title, and clicking it opens the drawer. Return with `control-openhands browser viewport desktop`.
+- **Phone drawer (`F02.mobile-drawer`).** Run `control-openhands browser viewport phone` and `control-openhands browser goto /conversations`. `control-openhands browser visible 'aside[data-collapsed]'` is `false`. Run `control-openhands browser click 'testid=sidebar-mobile-menu-toggle'` and `control-openhands browser wait 'testid=sidebar-mobile-drawer'`. `control-openhands browser bbox 'testid=sidebar-mobile-drawer'` gives width `300`, `insideViewport` `true` and `pageHorizontalOverflow` `false`, and `control-openhands browser screenshot --feature F02.mobile-drawer --name open` shows the full rail. Close paths, each followed by `control-openhands browser wait 'testid=sidebar-mobile-drawer' --state detached`: `control-openhands browser click 'testid=sidebar-mobile-drawer-close'`; `control-openhands browser press Escape`; `control-openhands browser click 'testid=sidebar-mobile-drawer >> testid=sidebar-automations-link'` (route change; the URL becomes `/automations`); and a backdrop click to the right of the drawer, `control-openhands browser mouse-click 350 400` (on `/conversations`, `control-openhands browser click 'testid=submit-button' --force` also lands on the backdrop, because the empty composer's send button sits at x≈326 under it). On a conversation page (`browser goto` the URL from `conversation start`), `control-openhands browser bbox 'testid=sidebar-mobile-menu-toggle'` is at the top-left of the chat header. `control-openhands browser screenshot --feature F02.mobile-drawer --name conversation-header` shows it beside the title, and clicking it opens the drawer. Modals over the drawer: on `/conversations` open the drawer, `control-openhands browser click 'testid=sidebar-mobile-drawer >> testid=conversation-panel-new-thread-picker'`, `control-openhands browser click 'text=+ Add Workspace'` (the `Add workspaces` dialog, `testid=folder-browser-modal`) and `control-openhands browser press Escape`: `browser count 'testid=folder-browser-modal'` is `0` and the drawer stays: `browser wait 'testid=sidebar-mobile-drawer' --state detached --timeout 3000` times out (an immediate count can still read `1` while a closing drawer animates out); a second Escape closes the drawer (the same wait without `--timeout` passes). The same holds for a row's Delete confirmation: hover `'testid=sidebar-mobile-drawer >> testid=conversation-card'`, click its `ellipsis-button`, then `testid=delete-button`, and press Escape: no `role=dialog` is left, the drawer and the conversation stay (`--name delete-confirm-after-escape`). Return with `control-openhands browser viewport desktop`.
- **Back chevron (`F02.mobile-top-bar`).** The Phone drawer bullet returned to desktop: run `control-openhands browser viewport phone`, `control-openhands browser goto /settings/llm` and `control-openhands browser attr 'testid=sidebar-mobile-back-button' aria-label` (`Settings`). Then run `control-openhands browser click 'testid=sidebar-mobile-back-button'` and `control-openhands browser url`; the URL is `/settings`, and on that page `browser count 'testid=sidebar-mobile-back-button'` is `0`. `control-openhands browser screenshot --feature F02.mobile-top-bar --name settings-root` shows only the hamburger above the settings menu. Do the same from `/skills`: the label is `Customize` and the click lands on `/customize`, which on phone is the hub (`extensions-mobile-hub`).
- **Phone command menu (`F02.command-menu-phone`).** At phone width open the drawer, then run `control-openhands browser click 'testid=sidebar-mobile-drawer >> testid=command-menu-trigger'`. `control-openhands browser bbox 'testid=command-menu >> role=listbox'` reports `insideViewport` `true` and `pageHorizontalOverflow` `false`, and `control-openhands browser screenshot --feature F02.command-menu-phone --name open` shows it. `control-openhands browser press Escape` closes only the menu: `browser count 'testid=command-menu'` is `0` and `control-openhands browser visible 'testid=sidebar-mobile-drawer'` is still `true`. Reopen it with `control-openhands browser click 'testid=sidebar-mobile-drawer >> testid=command-menu-trigger'`, type `secrets` and press Enter; `control-openhands browser wait-url 'secrets$'` passes and `control-openhands browser wait 'testid=sidebar-mobile-drawer' --state detached` passes.
- **Checklist card (`F02.checklist`).** The phone bullets above leave the phone viewport on `/settings/secrets`: run `control-openhands browser viewport desktop`, `control-openhands browser goto /conversations`, then `control-openhands browser count 'testid=sidebar-onboarding-checklist >> role=listitem'` (`6`) and `control-openhands browser text 'testid=sidebar-onboarding-checklist'`. The first lines are `Getting started` and `N complete`. `control-openhands browser snapshot 'aside[data-collapsed]'` shows the six links: `/settings/llm`, `/conversations`, `/automations`, `/settings/agents`, `/mcp` and `https://openhands.dev/joinslack`.
@@ -89,7 +89,7 @@ Preconditions:
- **Setup guide (`F02.super-admin-setup-guide`).** Local: `control-openhands browser count 'testid=super-admin-setup-guide'` is `0` at `desktop` and `phone`, `browser count 'testid=sidebar-onboarding-checklist'` is `1`, and `control-openhands browser network --filter setup-state` reports `total` `0`. The positive path is blocked locally: it needs an OHE backend with `ENABLE_SUPER_ADMIN` whose `/api/admin/setup-state` returns a `guide_org_id`, `guide_dismissed` `false` and at least one unfinished required step, signed in as the first Super Admin.
- **Setup tour (`F02.super-admin-setup-tour`).** Local, at `desktop` and `phone`: run `control-openhands browser network --clear`, `control-openhands browser goto '/?setup_tour=first-automation'`, wait about 6 s, then `control-openhands browser count '.oh-setup-tour-popover'` and `browser count 'testid=super-admin-setup-guide'` are `0`, `browser count 'testid=home-screen'` is `1`, `browser url` still ends in `?setup_tour=first-automation`, and `browser network --filter setup-state` reports `total` `0`. The tour itself is blocked with the same Enterprise prerequisite as the guide.
- **Update tile (`F02.update-tile`).** The tile is blocked in an up-to-date checkout or without registry access. Check with `control-openhands browser count 'testid=agent-canvas-version-tile'`. When the count is `1`, click it; the modal has `agent-canvas-update-command-npm`, `agent-canvas-update-command-docker` and `agent-canvas-update-check-button`. The manual check runs either way: on `/settings/app` (`control-openhands browser goto /settings/app`; the previous bullet ends on `/mcp`), `testid=agent-canvas-update-badge` reads `Up to date` (or `New version`), `control-openhands browser click 'testid=agent-canvas-update-toggle'` opens `agent-canvas-update-modal`, and `control-openhands browser click 'testid=agent-canvas-update-check-button'` reports the result in `agent-canvas-update-status` (`You're running the latest version. What's new` when current). Close it with `control-openhands browser click 'testid=close-agent-canvas-update-modal'`.
-- **Error toast (`F02.error-toasts`).** Run `control-openhands browser goto /conversations/00000000-0000-0000-0000-000000000001` and `control-openhands browser wait-text 'This conversation does not exist'`. The URL becomes `/conversations`, and `control-openhands browser screenshot --feature F02.error-toasts --name missing-conversation` shows a top-right error toast. It is the only toast: `control-openhands browser toasts` lists just the "This conversation does not exist, or you do not have permission to access it…" text, and `control-openhands browser toasts --history` has no entry containing `this UI does not understand`. `control-openhands browser wait 'text=This conversation does not exist' --state hidden` passes within 30 s. Light theme: arrange `control-openhands browser eval "localStorage.setItem('openhands-color-theme','light-plus')"`, repeat the `goto`, and read `control-openhands browser eval "(()=>{const t=[...document.querySelectorAll('[role=status]')].find(e=>e.textContent.includes('does not exist'));return getComputedStyle(t).color+' on '+getComputedStyle(t.closest('div[style]')).backgroundColor})()"`: `rgb(31, 31, 31) on rgb(229, 229, 229)`: every toast uses the theme's high-emphasis ink (`--oh-contrast`) on its surface (the default theme reads `rgb(255, 255, 255) on rgb(49, 49, 49)`). A success toast reads the same: save any change on `/settings/app` (for example `control-openhands browser click 'testid=app-settings-screen >> text=Sound Notifications'`, `control-openhands browser click 'testid=submit-button'`, then toggle it back and save again) and read the `Settings saved` toast the same way: `rgb(31, 31, 31) on rgb(229, 229, 229)`. Solarized Light (`solarized-light`) gives `rgb(0, 43, 54) on rgb(230, 223, 204)` for both toasts. Restore with `browser eval "localStorage.removeItem('openhands-color-theme')"`.
+- **Error toast (`F02.error-toasts`).** Run `control-openhands browser goto /conversations/00000000-0000-0000-0000-000000000001` and `control-openhands browser wait-text 'This conversation does not exist'`. The URL becomes `/conversations`, and `control-openhands browser screenshot --feature F02.error-toasts --name missing-conversation` shows a top-right error toast. It is the only toast: `control-openhands browser toasts` lists just the "This conversation does not exist, or you do not have permission to access it…" text, and `control-openhands browser toasts --history` has no entry containing `this UI does not understand`. `control-openhands browser wait 'text=This conversation does not exist' --state hidden` passes within 30 s. Light theme: arrange `control-openhands browser eval "localStorage.setItem('openhands-color-theme','light-plus')"`, repeat the `goto`, and read `control-openhands browser eval "(()=>{const t=[...document.querySelectorAll('[role=status]')].find(e=>e.textContent.includes('does not exist'));return getComputedStyle(t).color+' on '+getComputedStyle(t.closest('div[style]')).backgroundColor})()"`: `rgb(31, 31, 31) on rgb(229, 229, 229)`: every toast uses the theme's high-emphasis ink (`--oh-contrast`) on its surface (the default theme reads `rgb(255, 255, 255) on rgb(49, 49, 49)`). A success toast reads the same: save any change on `/settings/app` (for example `control-openhands browser click 'testid=app-settings-screen >> text=Sound Notifications'`, `control-openhands browser click 'testid=submit-button'`, then toggle it back and save again) and read the `Settings saved` toast the same way: `rgb(31, 31, 31) on rgb(229, 229, 229)`. Solarized Light (`solarized-light`) gives `rgb(0, 43, 54) on rgb(230, 223, 204)` for both toasts. The toast's icon follows the theme's status error colour: `control-openhands browser eval "getComputedStyle([...document.querySelectorAll('[role=status]')].find(e=>e.textContent.includes('does not exist')).closest('div[style]').querySelector('svg')).color"` is `rgb(209, 52, 56)` (`#D13438`) in Light+, `rgb(220, 50, 47)` (`#DC322F`) in Solarized Light and `rgb(255, 104, 78)` (`#FF684E`) in the default theme. Restore with `browser eval "localStorage.removeItem('openhands-color-theme')"`.
- **Check for errors.** Run `control-openhands browser errors --app-only` after each group; it reports `pageErrors` `0` and `appErrors` `0`. External `registry.npmjs.org` and `cdn.simpleicons.org` failures are sandbox noise.
- **Restore.** Leave the sidebar expanded, unpinned, with the checklist shown and expanded, at `browser viewport desktop`.
diff --git a/.agents/skills/verify-openhands/references/feature-map/F06-agent-activity.md b/.agents/skills/verify-openhands/references/feature-map/F06-agent-activity.md
index 01b3b9bb6c..878a75057b 100644
--- a/.agents/skills/verify-openhands/references/feature-map/F06-agent-activity.md
+++ b/.agents/skills/verify-openhands/references/feature-map/F06-agent-activity.md
@@ -47,7 +47,7 @@ Source: `src/components/features/chat/chat-interface.tsx`, `src/components/conve
- `F06.skill-install-banner`: after the agent installs a skill into the workspace, a banner offers "Start new conversation with this skill" (same workspace) and Close.
- `F06.llm-not-configured-banner`: with no usable LLM, a conversation shows "Your LLM isn't set up yet..." with Set up LLM; suggestions are hidden and sending is disabled.
- `F06.critic-result`: with the critic enabled, agent messages and the finishing action each carry a "Critic: agent success likelihood" block (stars labelled `Score: <N.N>%`, one decimal, with the percentage beside them); Expand details lists the issue categories (Potential Issues, Infrastructure, Likely Follow-up).
-- `F06.hook-events`: hook executions render as their own rows in the stream.
+- `F06.hook-events`: hook executions render as their own rows in the stream, each with an `ok`, `failed` or `blocked` badge; expanding a row shows its command, exit code, stderr, error or blocked reason and output, with the badges, stderr, error and blocked reason readable (at least 4.5:1) in dark and light themes.
- `F06.corrective-nudge`: when the model answers with neither a message nor a tool call, the SDK's nudge (`Your last response did not include a function call or a message. …`) shows as a muted, italic `role=note` line with an info icon (`corrective-nudge-message`), not as a user bubble (#17864).
- `F06.phone`: at 390 px the transcript fits the viewport; wide tables scroll inside their own container.
@@ -67,7 +67,7 @@ Preconditions:
- Baseline state (launched, doctored, `onboard --skip` done) and `control-openhands llm preset deepseek` (deepseek-flash active). Every prompt below is tiny and stays in the conversation's own workspace.
- Conversation ids are printed as `"id"` by `control-openhands conversation start`, or read from `control-openhands browser url` (`/conversations/<id>`). `<id>` below is always the conversation the step just created or opened.
- `F06.llm-not-configured-banner` needs a second, fresh `control-openhands launch --new --build never` with no `llm preset` (export its run dir as `OH_VERIFY_RUN` for those commands, then `control-openhands stop` it).
-- `F06.critic-result`: if an authorized OpenHands (`openhands/…`) or compatible OpenHands LLM-proxy profile is available, use it for the critic check, regardless of the profile used for the rest of the pass. The SDK reuses that profile's LLM key when Critic API Key is empty; a separate critic key is not required. For a non-default proxy deployment, use its compatible critic endpoint. Restore the previous profile after the check. Record `blocked` only when no authorized working critic configuration is available, not merely because the main profile uses another provider; its bullet also says how to arrange a locally scored block with a model only. The automatic-fetch half of `F06.load-older-history` (its 16-command conversation) needs a model; the scroll-to-top half is driven without one, on the dummy profile below, with thirty refused user messages. `F06.hook-events` needs a configured hook; none is set in a fresh run.
+- `F06.critic-result`: if an authorized OpenHands (`openhands/…`) or compatible OpenHands LLM-proxy profile is available, use it for the critic check, regardless of the profile used for the rest of the pass. The SDK reuses that profile's LLM key when Critic API Key is empty; a separate critic key is not required. For a non-default proxy deployment, use its compatible critic endpoint. Restore the previous profile after the check. Record `blocked` only when no authorized working critic configuration is available, not merely because the main profile uses another provider; its bullet also says how to arrange a locally scored block with a model only. The automatic-fetch half of `F06.load-older-history` (its 16-command conversation) needs a model; the scroll-to-top half is driven without one, on the dummy profile below, with thirty refused user messages. `F06.hook-events` needs a configured hook (none is set in a fresh run); its bullet arranges three in a fixture workspace and needs no model reply, because the last hook blocks the prompt.
- `F06.clock-skew-send` and `F06.failed-send-persists` need no model reply: a conversation that exists is enough (without a key the agent errors at once, and the user rows still land). Without a key, create the family's conversations with a dummy profile (`QA_DUMMY_KEY=qa control-openhands llm set --profile qa-zeta --model openai/gpt-4o --api-key-env QA_DUMMY_KEY --no-validate`) and `control-openhands conversation start --prompt ...` without `--wait`. `F06.corrective-nudge` needs a genuine provider turn with neither text nor a tool call; it is producer-dependent, not unavailable on Local. A confined two-iteration DeepSeek Flash attempt produced the real SDK nudge on 2026-10-17. Bound the attempt; never fabricate a response/event or guarantee an empty turn. Without one, record the missing producer precondition rather than a rendering pass or product failure.
- The harness browser grants clipboard access: `control-openhands browser clipboard` reads what a Copy button wrote and `browser clipboard --write qa-empty` resets it first.
- A terminal call can come back as "The terminal session was reset..." after this run's agent server restarts; the model reruns it, which turns one action into a group of two (see Gotchas).
@@ -86,7 +86,7 @@ Preconditions:
- **Empty state (`F06.empty-state-suggestions`).** Run `control-openhands browser click 'testid=conversation-panel-new-thread-picker'`, `control-openhands browser click 'testid=launch-no-workspace' --expect-url '/conversations/'`, then `control-openhands browser text 'testid=chat-suggestions'`: `Let's start building!` plus `Increase test coverage`, `Auto-merge PRs`, `Fix README`, `Clean dependencies`. Run `control-openhands browser click 'testid=chat-suggestions >> text=Fix README'` and `control-openhands browser text 'testid=chat-input'`: the composer holds the README-improvement prompt.
- **Pending bubble and long messages (`F06.pending-messages`, `F06.user-and-agent-messages`).** Write a 25-line message to a file (first line `Reply with only the word OK. Ignore the filler lines below.`, then `filler line 1`…`filler line 24`), run `control-openhands browser fill 'testid=chat-input' x --value-file <file>` and `control-openhands browser click 'testid=submit-button' --observe '[data-testid=chat-message-sending]' --observe-ms 3000`. `observed` shows `Sending...` then `<absent>` once the server echoes it. After `control-openhands conversation wait <id> --timeout 120`, `control-openhands browser testids 'testid=chat-scroll-container'` lists `chat-message-truncation-gradient` and `chat-message-expand` ("View More"); `control-openhands browser click 'testid=chat-message-expand'` removes both and the whole text shows. `testid=agent-message` reads `OK`.
- **Browser clock ahead (`F06.clock-skew-send`).** On the same conversation, skew the page's clock before loading it: `control-openhands browser clock --offset-ms 300000`, `control-openhands browser goto /conversations/<id>`, `control-openhands browser wait 'testid=chat-input'`; `control-openhands browser eval "new Date().toISOString()"` is about five minutes ahead of `date -u` (read-only shell check). Run `control-openhands browser fill 'testid=chat-input' 'Reply with only: clock-ok'` and `control-openhands browser click 'testid=submit-button' --observe '[data-testid=chat-message-sending],[data-testid=chat-message-error]' --observe-ms 6000`: `observed` shows `Sending...` and then `<absent>` (the server echo confirmed it), never `Failed to send`. `control-openhands browser count 'testid=chat-message-sending'` and `control-openhands browser count 'testid=chat-message-retry'` are `0`, and `control-openhands conversation events <id> --kinds MessageEvent` gains one user row `Reply with only: clock-ok` whose `ts` is the server's time (close to `date -u`), not the browser's. Undo the skew before moving on: `control-openhands browser clock --offset-ms 0` and `control-openhands browser reload` (with a model, `control-openhands conversation wait <id> --fresh --timeout 120` first).
-- **Server lost: banner, failed send, reconnect (`F06.error-banner`, `F06.pending-messages`, `F06.reconnect`).** With a conversation open run `control-openhands service stop agent-server` and `control-openhands browser wait 'testid=error-message-banner' --timeout 20000`: the banner reads `Unable to connect to server` with `error-message-banner-retry`, `-copy` and `-dismiss`; the composer shows `Disconnected`. Run `control-openhands browser fill 'testid=chat-input' 'Reply with only the word PONG.'` and `control-openhands browser click 'testid=submit-button' --observe '[data-testid=chat-message-sending],[data-testid=chat-message-error]' --observe-ms 8000`: `Sending...` turns into `Failed to send Retry Dismiss`. `control-openhands browser click 'testid=chat-message-retry'` while down fails again; `control-openhands browser click 'testid=chat-message-dismiss'` removes the bubble (`count 'testid=chat-message-error'` is `0`). Send the same message again so one failed bubble remains, `control-openhands browser click 'testid=error-message-banner-dismiss'` (banner count `0`), then `control-openhands restart`. Poll `control-openhands browser text 'testid=interactive-chat-box'` every 10 s: within about a minute after `restart` returns it no longer contains `Disconnected` (reconnected without reload). `control-openhands browser click 'testid=chat-message-retry'`, `control-openhands conversation wait <id> --timeout 120`, then `control-openhands conversation events <id> --kinds MessageEvent`: `Reply with only the word PONG.` appears once and the last `agent-message` is `PONG`.
+- **Server lost: banner, failed send, reconnect (`F06.error-banner`, `F06.pending-messages`, `F06.reconnect`).** With a conversation open run `control-openhands service stop agent-server` and `control-openhands browser wait 'testid=error-message-banner' --timeout 20000`: the banner reads `Unable to connect to server` with `error-message-banner-retry`, `-copy` and `-dismiss`, led by `error-message-banner-icon` in the theme's status error color (`browser eval` of its computed `color`: `rgb(255, 104, 78)` in the default theme, `rgb(209, 52, 56)` in Light+, at desktop and phone); the composer shows `Disconnected`. Run `control-openhands browser fill 'testid=chat-input' 'Reply with only the word PONG.'` and `control-openhands browser click 'testid=submit-button' --observe '[data-testid=chat-message-sending],[data-testid=chat-message-error]' --observe-ms 8000`: `Sending...` turns into `Failed to send Retry Dismiss`. `control-openhands browser click 'testid=chat-message-retry'` while down fails again; `control-openhands browser click 'testid=chat-message-dismiss'` removes the bubble (`count 'testid=chat-message-error'` is `0`). Send the same message again so one failed bubble remains, `control-openhands browser click 'testid=error-message-banner-dismiss'` (banner count `0`), then `control-openhands restart`. Poll `control-openhands browser text 'testid=interactive-chat-box'` every 10 s: within about a minute after `restart` returns it no longer contains `Disconnected` (reconnected without reload). `control-openhands browser click 'testid=chat-message-retry'`, `control-openhands conversation wait <id> --timeout 120`, then `control-openhands conversation events <id> --kinds MessageEvent`: `Reply with only the word PONG.` appears once and the last `agent-message` is `PONG`.
- **Banner Retry (`F06.error-banner`).** Run `control-openhands service stop agent-server`, wait for `testid=error-message-banner` as above, `control-openhands restart`, then `control-openhands browser click 'testid=error-message-banner-retry'`. `control-openhands browser count 'testid=error-message-banner'` is `0` and the composer is no longer Disconnected. Take `browser screenshot 'testid=chat-interface' --feature F06.error-banner --name disconnected` while the banner is up.
- **Failed send survives leaving (`F06.failed-send-persists`).** With `/conversations/<id>` open and healthy, run `control-openhands service stop agent-server`, `control-openhands browser wait 'testid=error-message-banner' --timeout 20000`, `control-openhands browser fill 'testid=chat-input' 'Reply with only: retry-ok'` and `control-openhands browser click 'testid=submit-button' --observe '[data-testid=chat-message-sending],[data-testid=chat-message-error]' --observe-ms 8000`: `Sending...` turns into `Failed to send Retry Dismiss` and `control-openhands browser count 'testid=chat-message-retry'` is `1` (`control-openhands browser screenshot 'testid=chat-interface' --feature F06.failed-send-persists --name failed-before-leaving`). Run `control-openhands restart` and `control-openhands browser wait 'testid=error-message-banner' --state hidden --timeout 90000` (the page reconnects by itself; `testid=error-message-banner-retry` hurries it while the banner is still up): the Retry row is still there (count `1`). Leave through the sidebar and come back: `control-openhands browser click 'role=link[name="New Chat"]'`, `control-openhands browser wait 'testid=home-chat-launcher'`, `control-openhands browser click 'a[href*="/conversations/<id>"] >> nth=0' --expect-url '/conversations/<id>'`, `control-openhands browser wait 'testid=user-message'`; `control-openhands browser count 'testid=chat-message-retry'` is still `1` after the history reloaded (`control-openhands browser screenshot 'testid=chat-interface' --feature F06.failed-send-persists --name retry-after-reopen`). Send the same text again (fill `Reply with only: retry-ok`, click `testid=submit-button` with the same `--observe`): the new bubble goes `Sending...` and settles while `Failed to send Retry Dismiss` stays; `control-openhands browser count 'testid=chat-message-sending'` is `0`, `control-openhands browser count 'testid=chat-message-retry'` is `1`, and the rows of `control-openhands conversation events <id> --kinds MessageEvent --grep 'retry-ok'` whose `source` is `user` number exactly one, `Reply with only: retry-ok` (in a model-free run `count` is `1`; with a model the agent's `retry-ok` reply matches the grep too, so count the `user` rows, not `count`). Then `control-openhands browser click 'testid=chat-message-retry'` and `control-openhands browser wait 'testid=chat-message-retry' --state detached --timeout 15000`: `control-openhands browser count 'testid=chat-message-sending'` is `0` and the same `--grep` now lists exactly two `user` rows `Reply with only: retry-ok` (one per attempt, none duplicated). With a model, `control-openhands conversation wait <id> --fresh --timeout 120` before the next bullet.
- **Live chip, stop and resume (`F06.live-activity`, `F06.stop-resume`).** Run `control-openhands conversation start --prompt "Run exactly one terminal command: sleep 25 && echo qa-slept. Then reply with only its output."` (no `--wait`), `control-openhands browser wait 'testid=live-activity-chip' --timeout 30000` and `control-openhands browser text 'testid=live-activity-chip'`: first `Thinking`, then the action title (e.g. `Run sleep 25 then echo qa-slept`); `browser text 'testid=interactive-chat-box'` contains `Running`. Run `control-openhands browser click 'testid=stop-button'`: `control-openhands conversation status <id>` is `paused`, the composer reads `Stopped`, `count 'testid=play-button'` is `1` and the chip is gone. Stop interrupts the running `sleep` too: `control-openhands conversation events <id>` shows an `AgentErrorEvent` (`Tool call interrupted before completion. The conversation was paused.`) followed by an `InterruptEvent`, and `browser text 'testid=chat-scroll-container'` shows an `Agent error` row under the action title. Run `control-openhands browser click 'testid=play-button'`: status `running`, then `control-openhands conversation wait <id> --timeout 120` ends `finished` with a new agent `MessageEvent` (the model either reports the interruption or reruns the command as a new action and replies `qa-slept`).
@@ -98,6 +98,7 @@ Preconditions:
- **Branch from here (`F06.branch-from-here`).** Open a conversation with at least two exchanges (the OK/PONG one above): `control-openhands browser goto /conversations/<id>` with its id (the previous bullet left `/settings/verification`). Agent message: `control-openhands browser hover 'testid=agent-message >> nth=0'`, `control-openhands browser click 'testid=agent-message >> nth=0 >> role=button[name="Branch from here"]' --expect-url '/conversations/(?!<id>)'`; after `control-openhands browser wait 'testid=user-message'` (the history mounts a moment after the URL changes, so an immediate count reads `0`) the new conversation has `count 'testid=user-message'` `1`, the agent message `OK`, and `browser eval "document.title"` contains `(branch)`. User message (edit): back on the OK/PONG conversation (`control-openhands browser goto /conversations/<id>`; the branch above has one user message), `control-openhands browser hover 'testid=user-message >> nth=1'` and `control-openhands browser click 'testid=user-message >> nth=1 >> role=button[name="Branch from here"]' --observe '[data-testid=chat-input]' --observe-ms 4000`: `observed` shows the composer change from empty to `Reply with only the word PONG.` and stay. The new conversation (`<branch-id>` from `browser url`) omits that message, `control-openhands browser text 'testid=chat-input'` is `Reply with only the word PONG.`, and nothing was sent: `control-openhands conversation events <branch-id> --kinds MessageEvent` lists only the copied first exchange (count `2`: the long message that starts `Reply with only the word OK.`, then `OK`; no `PONG`). `control-openhands browser enabled 'testid=submit-button'` should be `true` with the prefilled text; today it is `false` until the text is edited or the page is reloaded (see Gotchas). Send it with `control-openhands browser press Enter --selector 'testid=chat-input'` (Enter sends whatever the Send button's state is) and `control-openhands conversation wait <branch-id> --fresh --timeout 120`: expected `finished` with a reply; today it ends `error` (Agent Server bug, see Gotchas).
- **Images (`F06.image-attachments`).** Run `control-openhands fixture image --name qa-f06-img` (prints the PNG path), `control-openhands browser upload 'testid=upload-image-input' <png path>` (a thumbnail appears in the composer), fill `testid=chat-input` with `Reply with only the word SEEN.` and click `testid=submit-button`. The user bubble contains `image-carousel` / `image-preview`. Run `control-openhands browser hover 'testid=image-preview'`, `control-openhands browser click 'testid=expand-image-button'` (`count 'testid=image-lightbox'` is `1`), `control-openhands browser press Escape` (`0`), click `expand-image-button` again and `control-openhands browser click 'testid=image-lightbox-close'` (`0`).
- **Error events (`F06.error-events`).** Any run that ends in a `ConversationErrorEvent` (today: any message sent in a branched conversation) leaves `conversation wait <id>` at `"status": "error"`; `control-openhands browser testids 'testid=chat-interface'` lists `error-message-banner` with `warning-message-banner-icon` and the error's text, and the composer status reads `Error`. Read the code with `control-openhands api GET "/api/conversations/<id>/events/search?limit=3&sort_order=TIMESTAMP_DESC"`.
+- **Hook rows (`F06.hook-events`).** Arrange (outside the UI) `control-openhands fixture git-repo --name qa-hooks` and write `$OH_VERIFY_RUN/workspace/qa-hooks/.openhands/hooks.json` as `{"user_prompt_submit":[{"hooks":[{"type":"command","command":"echo qa-hook-stderr >&2; exit 0","timeout":10},{"type":"command","command":"sleep 2","timeout":1},{"type":"command","command":"echo '{\"decision\":\"deny\",\"reason\":\"QA_HOOK blocked\"}'; exit 2","timeout":10}]}]}`. Run `control-openhands conversation start --workspace qa-hooks --prompt "Reply with the single word: hi. Do not run any tools." --wait --timeout 180`: it finishes without an agent reply (the deny blocks the prompt) and `conversation events <id> --from-start` has three `HookExecutionEvent` rows. `control-openhands browser text 'testid=chat-interface'` lists `Hook: UserPromptSubmit` three times with the badges `ok`, `failed` and `blocked` (`browser count 'testid=generic-event-message-title'` is `3`). Expand each with `control-openhands browser click 'testid=generic-event-message-title >> nth=<0|1|2> >> role=button[name="Expand"]'`: the ok row shows `Command`, `Exit code: 0` and `Stderr: qa-hook-stderr`; the failed row `Exit code: -1` and `Error: Hook timed out after 1 seconds`; the blocked row `Exit code: 2`, `Blocked reason: QA_HOOK blocked` and the JSON under `Output`. The badges, the stderr box, the error and the blocked reason are readable in every theme: compute their text contrast against the composited background from `getComputedStyle` with `browser eval`, in the default theme, then Light+ and Solarized Light (switch on `/settings/app` as in F16, come back and expand again). Each is at least 4.5:1 (2026-10-19: default 4.92 to 9.54, Light+ 4.99 to 6.1, Solarized Light 4.78 to 6.5; phone 390×844 gives the Light+ values with the rows wrapping inside the width). `control-openhands browser screenshot 'testid=chat-scroll-container' --feature F06.hook-events --name <theme>` per theme. Restore OpenHands-Neutral afterwards.
- **Skill installed banner (`F06.skill-install-banner`).** The banner keys on the installer's success line in terminal output. Put this one line in a prompt file: `Run exactly this one terminal command, verbatim, then reply DONE: mkdir -p .agents/skills/qa-skill && printf '# qa-skill\n' > .agents/skills/qa-skill/SKILL.md && echo "✅ Successfully installed 'qa-skill' to $PWD/.agents/skills/qa-skill"`, then run `control-openhands conversation start --prompt "$(cat <prompt file>)" --wait --timeout 150`. `control-openhands browser text 'testid=skill-install-restart-banner'` reads `Installed to this workspace: qa-skill. Skills load when a conversation starts, so this conversation can't use them yet.` Run `control-openhands browser click 'testid=skill-install-restart-action' --expect-url '/conversations/(?!<id>)'`; `control-openhands api GET /api/conversations/<new id>` has the same `workspace.working_dir` as the source. Back on the source (`control-openhands browser goto /conversations/<id>`), `control-openhands browser click 'testid=skill-install-restart-dismiss'` hides it (count `0`); after `browser reload` it is back (dismissal is session-only by design).
- **No LLM (`F06.llm-not-configured-banner`).** On the fresh no-LLM run: `control-openhands onboard --skip`, `control-openhands browser click 'testid=conversation-panel-new-thread-picker'`, `control-openhands browser click 'testid=launch-no-workspace' --expect-url '/conversations/'`. `control-openhands browser text 'testid=home-llm-not-configured-banner'` reads `Your LLM isn't set up yet, so conversations won't run. Finish setup to get started. Set up LLM`; `count 'testid=chat-suggestions'` is `0`; the composer cannot take text (`control-openhands browser attr 'testid=chat-input' contenteditable` is `false`, so `browser fill` fails) and `control-openhands browser enabled 'testid=submit-button'` is `false`. `control-openhands browser click 'testid=home-llm-not-configured-action'` then `browser url` ends in `/settings/llm`.
- **Critic (`F06.critic-result`).** Toggle the hidden switch through its label on `/settings/verification`: `control-openhands browser goto /settings/verification`, `control-openhands browser click 'label:has([data-testid="sdk-settings-verification.critic_enabled"])'`, save, `api GET /api/settings` shows `"critic_enabled": true`. Use the authorized OpenHands/proxy-backed profile described above for this check; leave Critic API Key empty to reuse its LLM key. With that profile and the critic enabled, run `control-openhands conversation start --prompt "Create qa_critic.txt containing hi, then finish." --wait --timeout 180`: every scored event carries a block, so `control-openhands browser count 'text=Critic: agent success likelihood'` is at least `1` and a finishing action (`finish` tool row) shows its own block next to the agent's message; the stars carry the score as their label, which an ARIA snapshot does not list: `control-openhands browser count '[aria-label^="Score: "]'` is at least `2` and `control-openhands browser attr '[aria-label^="Score: "] >> nth=0' aria-label` reads `Score: <N.N>%` with one decimal (`Score: 82.0%`), the same number as the `(82.0%)` text beside the stars. `control-openhands browser click 'role=button[name="Expand details"] >> nth=0'` opens the categories `Potential Issues:`, `Infrastructure:` and `Likely Follow-up:` with their named issues in `control-openhands browser text 'testid=chat-scroll-container'`, and the button's label flips to `Collapse details`. Turn it off again the same way and restore the previous profile. Profile availability is a precondition, not proof that the service accepted the credential: require real scored events before recording a pass. A locally scored block needs no Cloud key: create a conversation through `control-openhands api POST /api/conversations --write --data @/dev/stdin`, piping a payload built from the environment (so the model key is neither printed nor written) with a direct `agent` `{kind:"Agent", llm:{model:"deepseek/deepseek-flash", api_key:<from $DEEPSEEK_API_KEY>}, tools:[], include_default_tools:["FinishTool"], critic:{kind:"AgentFinishedCritic"}}`, a `workspace.working_dir` under `$OH_VERIFY_RUN/workspace` and an `initial_message` (`run: true`) `Call the finish tool immediately with the message qa-critic-done. Do not do anything else.`; arrange, not proof. After `control-openhands conversation wait <id>` (`finished`) open `/conversations/<id>`: the block reads `Critic: agent success likelihood ☆☆☆☆☆ (0.0%)` with the iterative-refinement hint below, and the stars carry `aria-label` `Score: 0.0%`. Its colours follow the theme: header, `(0.0%)` and hint use tertiary text (8.45:1 on the Light+ chat panel, 4.99:1 on Solarized Light) and the score-0 stars use error feedback (5.86:1 and 5.61:1); this critic lists no issue categories, so Expand details is absent.
@@ -107,7 +108,7 @@ Preconditions:
- **Phone (`F06.phone`).** Run `control-openhands browser viewport phone`, open the table conversation (`control-openhands browser goto /conversations/<stream id>`), `control-openhands browser bbox 'testid=chat-interface'` (`insideViewport` `true`, `pageHorizontalOverflow` `false`) and `control-openhands browser bbox 'testid=markdown-table-scroll'` (`scrollWidth` larger than `clientWidth`: the table scrolls inside itself). `control-openhands browser screenshot --feature F06.phone --name conversation`, then `control-openhands browser viewport desktop`.
- After each group run `control-openhands browser errors --app-only`. Outages driven with `service stop` log connection-refused console errors by design; `pageErrors` must stay `0`.
-- **Planner discovery with many children (`F06.tool-visualizers`).** Expected: reloading a parent still discovers its `plannerparent`-tagged planner when it has 100 child conversations. Known failure (reproduced 2026-10-16, SDK 1.55.1, [OpenHands/software-agent-sdk#5790](https://github.com/OpenHands/software-agent-sdk/issues/5790)): the unchunked 100-ID request returns 500, shows `Internal Server Error` and omits the planner session connection. A99-child control returns 200 and opens both planner and parent sockets. These are genuine idle API-created children, without a model run or browser-storage changes; a new parent ID has no UI-created planner hint.
+- **Planner discovery with many children (`F06.tool-visualizers`).** Expected: reloading a parent still discovers its `plannerparent`-tagged planner when it has 100 child conversations. Known failure (reproduced 2026-10-19, SDK 1.55.1, [OpenHands/software-agent-sdk#5790](https://github.com/OpenHands/software-agent-sdk/issues/5790)): the unchunked 100-ID request returns 500, shows `Internal Server Error` and omits the planner session connection. A99-child control returns 200 and opens both planner and parent sockets. A newer Agent Server answers the 100-ID request with 422 (at most 99 IDs), so discovery also needs the UI to split the child-ID batch. These are genuine idle API-created children, without a model run or browser-storage changes; a new parent ID has no UI-created planner hint.
```sh known-bug F06.tool-visualizers planner-batch-100
control-openhands browser viewport desktop
QA_ID="$(python3 - <<'PY'
@@ -226,5 +227,4 @@ control-openhands assert --field value --equals inline -- browser eval 'getCompu
- Session reconnects use the latest received `after_seq` cursor (`F06.reconnect`). For a local metadata check, retain the open page, send genuine `run:false` user events, record `/events/count`, restart the owned backend and inspect finite WebSocket lifecycle metadata: the reconnect cursor is the previous count minus one and a subsequent recovery marker appears once. This does not measure frame-body replay or Cloud behavior.
- Stop persists `paused` before an in-flight model request completes (`F06.stop-resume`); the persisted state and reloaded Stopped UI survive an Agent Server restart. A bounded credential-free check can use the tracked accept-only endpoint and an exact listener PID proven to descend from the owned launcher; never kill by pattern or resume the paused conversation.
-- Known issue [#18521](https://github.com/OpenHands/OpenHands/issues/18521) (`F06.hook-events`): with real `user_prompt_submit` hooks in `<workspace>/.openhands/hooks.json` (stderr then exit 0; `sleep 2` with timeout 1; a JSON deny with exit 2), expanded hook rows in Light+ show the ok/failed/blocked badges at 1.84:1, 1.50:1 and 1.80:1, stderr at 1.30:1, the timeout error at 2.89:1 and the blocked reason at 1.72:1; Solarized Light gives 1.18:1 to 2.68:1.
- Known failure [software-agent-sdk#4554](https://github.com/OpenHands/software-agent-sdk/issues/4554) (`F06.critic-result`; reproduced 2026-10-17 with SDK 1.55.1): a genuine tracked README change through TerminalAction followed by Finish produces `AgentFinishedCritic` score zero. Expected behavior remains score one for a nonempty Git patch plus Finish; the installed `agent/critic_mixin.py` passes `git_patch=None`, preventing that condition from succeeding. The issue is closed, but its fix [PR #4585](https://github.com/OpenHands/software-agent-sdk/pull/4585) remains open/unmerged at this audit; do not retire the failure based on issue status. Separate genuine built-in `PassCritic` score-one events proved success text/stars rendering across five themes (and desktop/phone screenshots); this constant-score diagnostic neither classifies quality nor proves #4554 fixed. Rich intermediate/category probabilities and the settings-enabled Cloud path still require their genuine producer/authorized credentials. Retained live receipts: `weekly-intent-ui/evidence/intent-review/critic-source-identity.json`, `critic-positive-events.json`, `critic-git-diff.txt`, `passcritic-events.json`, `passcritic-success-colors.json` and `critic-hint-colors.json`; screenshots are under that run's `evidence/F06.critic-result/`. Never retain or publish credential-bearing request payloads.
diff --git a/.agents/skills/verify-openhands/references/feature-map/F07-conversation-page.md b/.agents/skills/verify-openhands/references/feature-map/F07-conversation-page.md
index 393a6cfff7..35288afeef 100644
--- a/.agents/skills/verify-openhands/references/feature-map/F07-conversation-page.md
+++ b/.agents/skills/verify-openhands/references/feature-map/F07-conversation-page.md
@@ -30,9 +30,9 @@ Source: `src/routes/conversation.tsx`, `src/components/features/conversation/`,
- `F07.delete-confirm`: Delete Conversation names the title in a confirmation; Cancel keeps it, Confirm Delete removes it and returns to `/conversations`.
- `F07.branch-from-message`: Branch from here on a message forks the conversation as `<title> (branch)`; on a user message the message is left out and its text goes back into the composer.
- `F07.right-panel-toggle`: the header panel button shows and hides the right tabs panel (closed again after a reload).
-- `F07.overview-toggle`: the header (i) button shows and hides the Overview column, which is exclusive with the right panel. At phone width it should reveal the Overview content on click. Known failure (reproduced twice 2026-10-17, [#16792](https://github.com/OpenHands/OpenHands/issues/16792)): at 390×844 the label changes from Show overview to Hide overview, but neither the Overview panel nor a phone dropdown renders.
+- `F07.overview-toggle`: the header (i) button shows and hides the Overview column, which is exclusive with the right panel. At phone width it should reveal the Overview content on click. Known failure (reproduced twice 2026-10-19, [#16792](https://github.com/OpenHands/OpenHands/issues/16792)): at 390×844 the label changes from Show overview to Hide overview, but neither the Overview panel nor a phone dropdown renders.
- `F07.git-actions-menu`: the header Git actions menu offers Commits, Pull, Push, Create PR and Create New Branch; each pre-fills the composer without sending, and Send becomes enabled so the user can submit or edit the prompt.
-- `F07.git-control-bar`: under the composer, a folder workspace shows an inert workspace chip and the branch; with a GitHub `origin` the repo and branch chips link to GitHub and Pull, Push and Pull Request appear.
+- `F07.git-control-bar`: under the composer, a folder workspace shows an inert workspace chip and the branch; with a GitHub `origin` the repo and branch chips link to GitHub and Pull, Push and Pull Request appear. The chip icons are visible in every theme. Known failure (reproduced twice 2026-10-19, [#18657](https://github.com/OpenHands/OpenHands/issues/18657)): in Light+ and Solarized Light the workspace (forked-repo) and branch chip icons are drawn white (`browser eval` of the `svg` computed `color` in `testid=chat-git-control-bar` is `rgb(255, 255, 255)`) on the light surface, so they are invisible at desktop and phone width; the default dark theme is unaffected.
- `F07.git-control-bar-send`: the Pull, Push and Pull Request chips send their git prompt to the agent at once.
- `F07.cloud-only`: Share publicly, Connect Repo, `task-<uuid>` start URLs, paused-sandbox resume, archived read-only state and the shared-conversation redirect exist only on a Cloud backend.
- `F07.phone`: at 390 px the header keeps its controls and the info dialogs fit the viewport.
diff --git a/.agents/skills/verify-openhands/references/feature-map/F08-workspace-files-and-changes.md b/.agents/skills/verify-openhands/references/feature-map/F08-workspace-files-and-changes.md
index f327646045..2982603f2c 100644
--- a/.agents/skills/verify-openhands/references/feature-map/F08-workspace-files-and-changes.md
+++ b/.agents/skills/verify-openhands/references/feature-map/F08-workspace-files-and-changes.md
@@ -95,7 +95,7 @@ Preconditions:
- **Diff modes (`F08.diff-view-modes`).** In the expanded Uncommitted row, run `control-openhands browser click 'testid=uncommitted-changes-row-content >> testid=file-diff-viewer-outer >> has-text=src/calc.py >> testid=collapse'`. `control-openhands browser text 'testid=uncommitted-changes-row-content >> testid=file-diff-viewer-outer >> has-text=src/calc.py >> testid=editor-container'` shows the added `def sub` lines, and `view-mode-diff` has `aria-pressed` `true`. The row's buttons are named: `control-openhands browser snapshot 'testid=uncommitted-changes-row-content >> testid=file-diff-viewer-outer >> has-text=src/calc.py'` lists `button "Old version"`, `button "Diff" [pressed]`, `button "New version"` and `button "Collapse"` (the same `collapse` button reads `Expand` while the row is collapsed), so `… >> role=button[name="Old version"]` works as well as the test id. Click `… >> testid=view-mode-old` (2 lines), then `… >> testid=view-mode-new` (4 lines), then `view-mode-diff`. Run `control-openhands browser errors --app-only` before and after: each collapse and each diff→old switch adds `TextModel got disposed before DiffEditorWidget model got reset` (fail, #17567).
- **Markdown preview (`F08.diff-markdown-preview`).** In this order `qa-notes.md` is already committed, so check it in the `QA agent commit` row: `control-openhands browser click 'testid=commit-row >> has-text=QA agent commit >> testid=commit-row-toggle'`, then with `F='testid=commit-row >> has-text=QA agent commit >> testid=file-diff-viewer-outer >> has-text=qa-notes.md'` click `"$F >> testid=collapse"` and `"$F >> testid=view-mode-new"`. `control-openhands browser count "$F >> testid=markdown-preview"` is `1`, rendering `<h1>QA notes</h1>`, **world** in bold and the `edited by agent` list item. (Before the second prompt the same check works under `testid=uncommitted-changes-row-content`.) Re-expand Uncommitted with `uncommitted-changes-row-toggle` for the next bullet.
- **Deleted file (`F08.diff-deleted-file`).** Expand `src/test_calc.py` in Uncommitted. `control-openhands browser text 'testid=uncommitted-changes-row-content >> testid=file-diff-viewer-outer >> has-text=src/test_calc.py >> testid=file-deleted-message'` reads `This file was deleted. Its previous contents are no longer available for preview.`
-- **From the overview (`F08.changes-from-overview`).** With the drawer closed, run `control-openhands browser click 'testid=conversation-overview-toggle'` and `control-openhands browser click 'testid=conversation-overview-diffs'` (its siblings `conversation-overview-diffs-additions`/`-deletions` show `+N`/`-N`; deleted files are not counted). `right-panel-toggle` `aria-pressed` is `true`, `testid=commit-list` is visible, `uncommitted-changes-row-toggle` has `aria-expanded` `true` and `testid=conversation-overview-panel` is hidden. After `browser reload` and the toggle, Commits is still selected.
+- **From the overview (`F08.changes-from-overview`).** With the drawer closed, run `control-openhands browser click 'testid=conversation-overview-toggle'` and `control-openhands browser click 'testid=conversation-overview-diffs'` (its siblings `conversation-overview-diffs-additions`/`-deletions` show `+N`/`-N` in the theme's status success and error colors, for example `rgb(31, 189, 83)`/`rgb(255, 104, 78)` in the default theme and `rgb(22, 130, 93)`/`rgb(209, 52, 56)` in Light+; deleted files are not counted). `right-panel-toggle` `aria-pressed` is `true`, `testid=commit-list` is visible, `uncommitted-changes-row-toggle` has `aria-expanded` `true` and `testid=conversation-overview-panel` is hidden. After `browser reload` and the toggle, Commits is still selected.
- **Empty states (`F08.commits-states`, `F08.files-empty-workspace`).** Run `control-openhands fixture folder`, which makes `<run>/workspace/qa-folder`, a plain folder (no `.git`) with `notes.md`; by the time the conversation has run it holds an empty `.git` (seen live; not created by the prompt), so it ends up a repo with no commits. Run `control-openhands browser goto /` (the browser is on the qa-repo conversation) and pick it with the workspace picker as in Preconditions (`folder-browser-entry-qa-folder`). Then run `control-openhands conversation start --stay --prompt 'Run exactly this bash command, then reply only "done": rm notes.md && git status --short' --wait --timeout 180`. Open the drawer and click `testid=conversation-tab-commits`; `control-openhands browser count 'text="No commits yet"'` is `1` (`browser screenshot --feature F08.commits-states --name no-commits`). On Files, click `testid=files-tab-refresh`; `control-openhands browser text 'testid=files-tab-tree'` reads `No files in workspace`. `Loading changes...` (`testid=commits-tab-status`) shows briefly on the first Commits open after a load: on `/conversations/<id>` with Files selected, `browser reload`, open the drawer, then `control-openhands browser click 'testid=conversation-tab-commits' --observe 'testid=commits-tab-status' --observe-ms 4000` records `Loading changes...` for a few ms. `Waiting for runtime to start...` (agent state init/loading/error) was not reached: `service stop agent-server` and `restart` both left `No commits yet` on screen.
- **Commit cap (`F08.commits-cap`).** Arrange more than 50 commits in the fixture (read-only for the UI): `for i in $(seq 1 50); do git -C "$OH_VERIFY_RUN/workspace/qa-repo" -c user.name=QA-Bulk -c [email protected] commit -q --allow-empty -m "QA bulk $i"; done`. On the qa-repo conversation (`control-openhands browser goto /conversations/<id>`; Empty states left the qa-folder conversation) run `browser reload`, open the drawer on Commits and wait for `testid=commit-list`. `control-openhands browser text 'testid=commit-list-cap-notice'` reads `Showing the latest 50 commits` and `browser count 'testid=commit-row'` is `50`. Do this last: it changes the commit list the other bullets read.
- **Phone panel page (`F08.phone-panel-page`).** Run `control-openhands browser viewport phone` on `/conversations/<id>`, then `control-openhands browser click 'testid=right-panel-toggle' --expect-url '/panel$'`. `control-openhands browser bbox 'testid=files-tab'` gives `insideViewport` `true` and `pageHorizontalOverflow` `false`, with Commits diffs usable at 390 px (`browser screenshot --feature F08.phone-panel-page --name panel`). `control-openhands browser attr 'testid=conversation-mobile-panel-back' aria-label` is `Back`. Run `control-openhands browser click 'testid=conversation-mobile-panel-back' --expect-url '/conversations/[^/]+$'`; the chat is visible. Return with `control-openhands browser viewport desktop`.
@@ -106,7 +106,7 @@ Preconditions:
- **Git filename with a space (`F08.uncommitted-changes`).** In a disposable workspace Git fixture, create and commit `my notes.md`, then modify it (plain shell Git/file commands are fixture setup, not UI proof). Open its empty Local conversation's Commits tab and read `browser network --filter '/git/'`. Expected: the modified path and other changes stay listed. Known failure (reproduced 2026-10-12, [OpenHands/software-agent-sdk#5642](https://github.com/OpenHands/software-agent-sdk/issues/5642)): the scoped changes call returns 400 and hides the Uncommitted changes. Only the space-name case was tested; the same issue's non-ASCII and rename cases still require their own checks.
- **File listing above one MiB (`F08.files-tree`).** In a disposable fixture, arrange 1800 tiny files beneath three 220-character directory segments so the sorted relative-path listing exceeds one MiB while remaining below the 2000-file cap. Keep a normal tracked README at the root. Open Files, click `testid=files-tab-refresh`, then count `role=button[name="README.md"][exact]` and inspect the tree. Known failure (reproduced 2026-10-12, [OpenHands/software-agent-sdk#5638](https://github.com/OpenHands/software-agent-sdk/issues/5638)): the README disappears and a spurious partial-path folder appears because the command API returns the tail output chunk only. The greater-than-100-chunk/null-exit case was not driven.
-- **Subdirectory Git paths (`F08.uncommitted-changes`, `F08.diff-view-modes`).** Expected: a workspace at `repo/src` lists changes beneath that folder relative to it, and its diffs open. Known failure (reproduced 2026-10-16, SDK 1.55.0, [OpenHands/software-agent-sdk#5843](https://github.com/OpenHands/software-agent-sdk/issues/5843)): tracked paths retain the repository prefix while untracked paths are relative to `src`; opening displayed `src/app.py` requests `repo/src/src/app.py` and returns 400. The ordinary repository-root control lists correct paths. The recipe creates only disposable, idle offline conversations.
+- **Subdirectory Git paths (`F08.uncommitted-changes`, `F08.diff-view-modes`).** Expected: a workspace at `repo/src` lists changes beneath that folder relative to it, and its diffs open. Known failure (reproduced 2026-10-19, SDK 1.55.1, [OpenHands/software-agent-sdk#5843](https://github.com/OpenHands/software-agent-sdk/issues/5843)): tracked paths retain the repository prefix while untracked paths are relative to `src`; opening displayed `src/app.py` requests `repo/src/src/app.py` and returns 400. The ordinary repository-root control lists correct paths. The recipe creates only disposable, idle offline conversations.
```sh known-bug F08.diff-view-modes subdirectory-paths
control-openhands browser viewport desktop
QA_ID="$(python3 - <<'PY'
diff --git a/.agents/skills/verify-openhands/references/feature-map/F19-plugins.md b/.agents/skills/verify-openhands/references/feature-map/F19-plugins.md
index 0d5994a2b5..bf9226749f 100644
--- a/.agents/skills/verify-openhands/references/feature-map/F19-plugins.md
+++ b/.agents/skills/verify-openhands/references/feature-map/F19-plugins.md
@@ -31,7 +31,7 @@ Source: `src/routes/skills-plugins.tsx`, `src/routes/launch.tsx`, `src/component
- `F19.local-plugin`: a plugin found in `~/.agents/plugins` shows a **Local** badge (no Install or switch), is listed under the Local filter, and its modal has a Local pill, Files, no Install/Uninstall and no Start Conversation.
- `F19.start-conversation`: **Start Conversation** in the detail modal opens `/launch?plugins=<base64 JSON>` for that plugin.
- `F19.launch-review`: `/launch` shows a modal titled `Launch <plugin>` (one plugin) or `Launch Plugin` (several), the optional `message` with HTML stripped and capped at 500 characters, and each plugin as `<source> / <path> @ <ref>`; X and Escape go to `/conversations`.
-- `F19.launch-parameters`: plugins with `parameters` get an expanded, collapsible section with a text input, number input (blank becomes0) or checkbox per value; edits are retained and accepted configuration must be honored through a supported backend path or clearly rejected as unsupported. Known failure (reproduced twice 2026-10-17, [#18614](https://github.com/OpenHands/OpenHands/issues/18614)): Local creation accepts edited ASCII values but sends only plugin source/ref/repo_path, silently dropping parameters; both actual requests returned 201. The current SDK 1.55.1 PluginSource schema does not accept parameters, so this does not claim simply forwarding the field would implement runtime support.
+- `F19.launch-parameters`: plugins with `parameters` get an expanded, collapsible section with a text input, number input (blank becomes 0) or checkbox per value, and edits are retained. The Local backend does not support plugin parameters: while any plugin in the review still carries a non-empty `parameters` object, an alert (`testid=plugin-parameters-unsupported`) reads `This local backend does not support plugin parameters. Remove the parameters from the launch link before starting a conversation.`, **Start Conversation** stays disabled even after the trust box is ticked, and no conversation-create request is sent. Plugins without parameters (or with an empty `parameters` object) launch normally.
- `F19.launch-trust`: **Start Conversation** stays disabled until the trust checkbox (`I trust this plugin from <sources> …`) is ticked; then it creates the conversation (sending the message, if any), opens `/conversations/<id>`, and the plugin is loaded there.
- `F19.launch-dev-params`: `/launch?plugin_source=…&plugin_ref=…&plugin_repo_path=…&message=…` opens the same review for one plugin.
- `F19.launch-errors`: bare `/launch` or an empty array says `No plugins were specified…`; bad base64, invalid JSON, a non-array or an item without `source` says `The plugin configuration is invalid…`; **Go Home** opens `/conversations`.
@@ -79,7 +79,7 @@ Preconditions:
- **Local plugin (`F19.local-plugin`).** After arranging `qa-local-plugin` (Preconditions), run `control-openhands browser reload`, `control-openhands browser text 'testid=plugin-local-badge-qa-local-plugin'` (`Local`), and `control-openhands browser count` on `testid=plugin-install-qa-local-plugin`, `testid=plugin-toggle-qa-local-plugin` and `testid=plugin-source-qa-local-plugin` (all `0`). `control-openhands browser click 'testid=plugins-filter-local'` leaves one card; under `plugins-filter-available` `browser count 'testid=plugin-card-qa-local-plugin'` is `0`. Back on All, `control-openhands browser click 'testid=plugin-card-qa-local-plugin'` and `control-openhands browser snapshot 'testid=plugin-detail-modal'`: no source line, pills `Local v1.0.0`, the bundled skill and Files, and only a **Close** button in the footer (`browser screenshot --feature F19.local-plugin --name modal`). Close with Escape. Clean up: `rm -rf "$OH_VERIFY_RUN/private/home/.agents/plugins/qa-local-plugin"`, `control-openhands browser reload`; the card count for `qa-local-plugin` is `0`.
- **Start a conversation from the catalog (`F19.start-conversation`, `F19.launch-trust`).** With `magic-test` not installed, run `control-openhands browser click 'testid=plugin-card-magic-test'`, `control-openhands browser click 'testid=plugin-detail-start-conversation-magic-test' --expect-url '/launch\?plugins='` and the `atob` eval above: one object with the catalog `source`, `ref: null`, `repo_path: null`. `control-openhands browser snapshot 'testid=plugin-launch-modal'` shows `Launch magic-test`, the source path and the checkbox `I trust this plugin from <source path> with the agent secrets defined in my account.`; `control-openhands browser enabled 'testid=start-conversation-button'` is `false`. Run `control-openhands browser check 'testid=trust-checkbox'` (`enabled` turns `true`) and `control-openhands browser click 'testid=start-conversation-button' --expect-url '/conversations/[0-9a-f-]+'`; `control-openhands browser url` gives `<id>`. `control-openhands browser click 'testid=chat-plus-button'`, `control-openhands browser click 'testid=show-plugins-button'` and `control-openhands browser text 'testid=plugins-modal'` list `magic-test`; close with `testid=close-plugins-modal`. Then `control-openhands browser type 'testid=chat-input' 'alakazam'`, `control-openhands browser press Enter --selector 'testid=chat-input'`, `control-openhands conversation wait <id> --timeout 180` and `control-openhands browser text 'testid=agent-message >> nth=-1'`: the magic phrase.
- **Review screen (`F19.launch-review`).** Run `control-openhands browser goto '/launch?plugins=W3sic291cmNlIjoiZ2l0aHViOk9wZW5IYW5kcy9leHRlbnNpb25zIiwicmVwb19wYXRoIjoicGx1Z2lucy9jaXR5LXdlYXRoZXIifSx7InNvdXJjZSI6Imh0dHBzOi8vZ2l0aHViLmNvbS9PcGVuSGFuZHMvZXh0ZW5zaW9ucy5naXQiLCJyZXBvX3BhdGgiOiJwbHVnaW5zL21hZ2ljLXRlc3QiLCJyZWYiOiJtYWluIn1d'` (two plugins from `OpenHands/extensions`, one as an HTTPS git URL) and `control-openhands browser snapshot 'testid=plugin-launch-modal'`: the heading `Launch Plugin` (`Launch` once); under `Plugins`, the text `city-weather OpenHands/extensions/ plugins/city-weather magic-test OpenHands/extensions/ plugins/magic-test@ main` (the screenshot renders `OpenHands/extensions / plugins/magic-test @ main`); and the checkbox `I trust this plugin from OpenHands/extensions with the agent secrets defined in my account.`, which names `OpenHands/extensions` once (`browser screenshot --feature F19.launch-review --name multi`). Message: `control-openhands browser goto '/launch?plugin_source=github%3AOpenHands%2Fextensions&message=%3Cb%3EWeather%3C%2Fb%3E%20for%20QA'` and `control-openhands browser text 'testid=plugin-launch-modal >> p'`: `Weather for QA` (tags stripped). With a 600-character `message` (`/launch?plugin_source=github%3AOpenHands%2Fextensions&message=aaa…`), `control-openhands browser eval "document.querySelector('[data-testid=plugin-launch-modal] p').textContent.length"` is `500`. Close: `control-openhands browser click 'testid=plugin-launch-modal >> testid=close-button' --expect-url '/conversations(\?|$)'`; on a fresh `/launch?...`, `control-openhands browser press Escape` also lands on `/conversations`.
-- **Parameters (`F19.launch-parameters`).** Run `control-openhands browser goto '/launch?plugins=W3sic291cmNlIjoiZ2l0aHViOk9wZW5IYW5kcy9leHRlbnNpb25zIiwicmVwb19wYXRoIjoicGx1Z2lucy9jaXR5LXdlYXRoZXIiLCJyZWYiOiJtYWluIiwicGFyYW1ldGVycyI6eyJjaXR5IjoiUGFyaXMiLCJkYXlzIjozLCJtZXRyaWMiOnRydWV9fV0%3D'` (city-weather with `{"city":"Paris","days":3,"metric":true}`) and `control-openhands browser snapshot 'testid=plugin-launch-modal'`: a `city-weather` section button, `ref: main`, `path: plugins/city-weather`, textbox `city` (`Paris`), spinbutton `days` (`3`) and a checked checkbox `metric`. `control-openhands browser click 'testid=plugin-section-0'` collapses it (`browser count 'testid=plugin-0-param-city'` is `0`); click again to expand. `control-openhands browser fill 'testid=plugin-0-param-city' Berlin`, `control-openhands browser fill 'testid=plugin-0-param-days' ''` (then `control-openhands browser value 'testid=plugin-0-param-days'` is `0`) and `control-openhands browser uncheck 'testid=plugin-0-param-metric'` are accepted. A link that mixes plugins with and without parameters (`/launch?plugins=W3sic291cmNlIjoiZ2l0aHViOk9wZW5IYW5kcy9leHRlbnNpb25zIiwicmVwb19wYXRoIjoicGx1Z2lucy9tYWdpYy10ZXN0In0seyJzb3VyY2UiOiJnaXRodWI6T3BlbkhhbmRzL2V4dGVuc2lvbnMiLCJyZXBvX3BhdGgiOiJwbHVnaW5zL2NpdHktd2VhdGhlciIsInBhcmFtZXRlcnMiOnsiY2l0eSI6IlBhcmlzIn19XQ%3D%3D`) lists the parameter-less one under `Additional Plugins`. Leave with Escape; do not start this one (see Gotchas).
+- **Parameters (`F19.launch-parameters`).** Run `control-openhands browser goto '/launch?plugins=W3sic291cmNlIjoiZ2l0aHViOk9wZW5IYW5kcy9leHRlbnNpb25zIiwicmVwb19wYXRoIjoicGx1Z2lucy9jaXR5LXdlYXRoZXIiLCJyZWYiOiJtYWluIiwicGFyYW1ldGVycyI6eyJjaXR5IjoiUGFyaXMiLCJkYXlzIjozLCJtZXRyaWMiOnRydWV9fV0%3D'` (city-weather with `{"city":"Paris","days":3,"metric":true}`) and `control-openhands browser snapshot 'testid=plugin-launch-modal'`: a `city-weather` section button, `ref: main`, `path: plugins/city-weather`, textbox `city` (`Paris`), spinbutton `days` (`3`) and a checked checkbox `metric`. `control-openhands browser click 'testid=plugin-section-0'` collapses it (`browser count 'testid=plugin-0-param-city'` is `0`); click again to expand. `control-openhands browser fill 'testid=plugin-0-param-city' Berlin`, `control-openhands browser fill 'testid=plugin-0-param-days' ''` (then `control-openhands browser value 'testid=plugin-0-param-days'` is `0`) and `control-openhands browser uncheck 'testid=plugin-0-param-metric'` are accepted. The alert stays visible whether the section is open or collapsed: `control-openhands browser text 'testid=plugin-parameters-unsupported'` reads `This local backend does not support plugin parameters. Remove the parameters from the launch link before starting a conversation.` Run `control-openhands browser check 'testid=trust-checkbox'`; `control-openhands browser enabled 'testid=start-conversation-button'` stays `false`. `control-openhands browser network --clear`, `control-openhands browser click 'testid=start-conversation-button' --force` and `control-openhands browser network --filter '/api/conversations$'` (`total` `0`): nothing is sent and the URL stays on `/launch` (`browser screenshot --feature F19.launch-parameters --name desktop-unsupported`). A link that mixes plugins with and without parameters (`/launch?plugins=W3sic291cmNlIjoiZ2l0aHViOk9wZW5IYW5kcy9leHRlbnNpb25zIiwicmVwb19wYXRoIjoicGx1Z2lucy9tYWdpYy10ZXN0In0seyJzb3VyY2UiOiJnaXRodWI6T3BlbkhhbmRzL2V4dGVuc2lvbnMiLCJyZXBvX3BhdGgiOiJwbHVnaW5zL2NpdHktd2VhdGhlciIsInBhcmFtZXRlcnMiOnsiY2l0eSI6IlBhcmlzIn19XQ%3D%3D`) lists the parameter-less one under `Additional Plugins`, shows the same alert and keeps Start disabled after trust. A link whose only plugin has `"parameters":{}` (`/launch?plugins=W3sic291cmNlIjoiZ2l0aHViOk9wZW5IYW5kcy9leHRlbnNpb25zIiwicmVwb19wYXRoIjoicGx1Z2lucy9tYWdpYy10ZXN0IiwicGFyYW1ldGVycyI6e319XQ%3D%3D`) shows no alert (`browser count 'testid=plugin-parameters-unsupported'` is `0`) and Start enables after trust. Leave each with Escape. At phone width (`browser viewport phone`) the first link gives the same alert and disabled Start, with `browser bbox 'testid=plugin-parameters-unsupported'` `insideViewport` `true` (`browser screenshot --feature F19.launch-parameters --name phone-unsupported`); return with `browser viewport desktop`.
- **Dev format (`F19.launch-dev-params`).** Run `control-openhands browser goto '/launch?plugin_source=github%3AOpenHands%2Fextensions&plugin_ref=main&plugin_repo_path=plugins%2Fmagic-test&message=hi%20QA'` and `control-openhands browser snapshot 'testid=plugin-launch-modal'`: `Launch magic-test`, paragraph `hi QA`, text `magic-test OpenHands/extensions/ plugins/magic-test@ main`.
- **Error states (`F19.launch-errors`).** For each of `/launch`, `/launch?plugins=garbage`, `/launch?plugins=e30%3D` (`{}`), `/launch?plugins=W10%3D` (`[]`) and `/launch?plugins=W3sicmVmIjoibWFpbiJ9XQ%3D%3D` (`[{"ref":"main"}]`) run `control-openhands browser goto '<url>'` and `control-openhands browser text 'testid=launch-error'`. Bare and `[]` read `Unable to Launch` / `No plugins were specified. Please provide at least one plugin to launch.`; the other three read `The plugin configuration is invalid. Please check the URL and try again.` (`browser screenshot --feature F19.launch-errors --name invalid-format`). `control-openhands browser count 'testid=try-again-button'` is `0` in all of them. `control-openhands browser click 'testid=go-home-button' --expect-url '/conversations(\?|$)'` leaves. The `Unable to Launch` heading stays at least 4.5:1 against the error card in all five themes at desktop and phone width (it uses the theme's error colour).
- **Creation failure (`F19.launch-creation-failed`).** Run `control-openhands browser goto '/launch?plugin_source=github%3AOpenHands%2Fextensions&plugin_repo_path=plugins%2Fmagic-test'`, `control-openhands browser check 'testid=trust-checkbox'`, `control-openhands service stop agent-server`, `control-openhands browser click 'testid=start-conversation-button'`, `control-openhands browser wait 'testid=launch-error' --timeout 60000` and `control-openhands browser text 'testid=launch-error'`: `Unable to Launch`, `Failed to Start Conversation`, `Go Home`, `Try Again` (`browser screenshot --feature F19.launch-creation-failed --name error`). Heading and **Try Again** stay at least 4.5:1 against their backgrounds, Try Again also on hover (checked in Solarized Light at desktop and phone width; Try Again uses the theme's contrast colours). `control-openhands browser click 'testid=try-again-button'` reloads the page; with the server still down the app shows **Manage backends** with the Local backend `Disconnected`. Restore with `control-openhands restart --timeout 240` and `control-openhands browser reload`: the review modal is back and `control-openhands browser eval "document.querySelector('[data-testid=trust-checkbox]').checked"` is `false`. Run `control-openhands doctor` before continuing.
@@ -99,7 +99,7 @@ Preconditions:
- **Disabled plugin reconciles on resume (`F19.enabled-autoload`).** Arrange an own installed `qa-resume-plugin` with a valid `.plugin/plugin.json`, `commands/qa-hello.md`, `skills/qa-plugin-skill/SKILL.md` and a `UserPromptSubmit` hook containing only `echo QA_HOOK`. Start a real Home chat with the dummy loopback profile; inspect its HookExecutionEvent (one) and plugin skill names through `api GET '/api/conversations/<id>?include_skills=true' --pick agent.agent_context.skills` (keep the full response private). Disable the real `plugin-toggle-qa-resume-plugin` with `--hover-first 250`, verify `GET /api/plugins/installed` says false, leave Plugins, `restart`, `doctor`, reload the conversation and `conversation send <id> --prompt '/qa-resume-plugin:qa-hello'`. Expected: the disabled command does not activate and plugin skills disappear. Known failure, reproduced in two independent persisted conversations on 2026-10-12 ([OpenHands/software-agent-sdk#5628](https://github.com/OpenHands/software-agent-sdk/issues/5628)): the resumed user MessageEvent activated `qa-resume-plugin:qa-hello`, and both plugin skills remained; the original hook count stayed one, so hooks did follow disable. A fresh chat begun while disabled activated no command and ran no hook. These are real pre-model activation/state observations; no successful reply or plugin MCP reconciliation is claimed. Uninstall only the own plugin and remove its source fixture afterwards.
-- **Unavailable plugin source is a handled Update error (`F19.refresh`).** Expected:400 for a vanished source, with the installation unchanged. Known failure (reproduced twice 2026-10-16, [OpenHands/software-agent-sdk#5897](https://github.com/OpenHands/software-agent-sdk/issues/5897)): two real Update clicks after moving only the owned source aside showed the missing-path error and returned 500; installed version/enabled state stayed unchanged. This is the plugin branch; Canvas has no skill-management Refresh action.
+- **Unavailable plugin source is a handled Update error (`F19.refresh`).** Expected:400 for a vanished source, with the installation unchanged. Known failure (reproduced twice 2026-10-19, [OpenHands/software-agent-sdk#5897](https://github.com/OpenHands/software-agent-sdk/issues/5897)): two real Update clicks after moving only the owned source aside toasted `Unable to parse extension source: local. Expected formats: 'github:owner/repo', git URL, or local path` and returned 500; installed version/enabled state stayed unchanged. This is the plugin branch; Canvas has no skill-management Refresh action.
```sh known-bug F19.refresh missing-source
QA_PLUGIN_SOURCE="$OH_VERIFY_RUN/workspace/qa-known-refresh-plugin"
@@ -120,14 +120,13 @@ Preconditions:
control-openhands assert --field recent.0.status --equals 400 --bug 'recent.0.status eq 500 && recent.0.path eq "/api/plugins/installed/qa-known-refresh-plugin/refresh"' -- browser network --filter '/api/plugins/installed/qa-known-refresh-plugin/refresh' # bug
```
-This block checks the accepted-create path: a 201 response must preserve the edited values. An explicit unsupported-configuration rejection is a separate acceptable product outcome and would require reviewing this recipe before treating an xpass as a fix.
+- **Local rejection with an owned fixture (`F19.launch-parameters`).** The same check on an inert run-owned plugin, asserting the alert, the disabled Start and that no create request is sent:
-```sh known-bug F19.launch-parameters local-delivery
-QA_PARAM_SOURCE="$OH_VERIFY_RUN/workspace/qa-known-parameter-plugin"
+```sh
+QA_PARAM_SOURCE="$OH_VERIFY_RUN/workspace/qa-parameter-plugin"
mkdir -p "$QA_PARAM_SOURCE/.plugin"
-printf '%s\n' '{"name":"qa-known-parameter-plugin","version":"0.1.0","description":"Inert owned parameter delivery fixture; no skills, tools or code."}' > "$QA_PARAM_SOURCE/.plugin/plugin.json"
-QA_CREATED_ID=''
-trap 'if [ -n "$QA_CREATED_ID" ]; then control-openhands api DELETE "/api/conversations/$QA_CREATED_ID" --write >/dev/null; fi; rm -rf "$QA_PARAM_SOURCE"; control-openhands browser goto /conversations >/dev/null' EXIT
+printf '%s\n' '{"name":"qa-parameter-plugin","version":"0.1.0","description":"Inert owned parameter fixture; no skills, tools or code."}' > "$QA_PARAM_SOURCE/.plugin/plugin.json"
+trap 'rm -rf "$QA_PARAM_SOURCE"; control-openhands browser goto /conversations >/dev/null' EXIT
QA_PARAM_PAYLOAD=$(jq -nc --arg source "$QA_PARAM_SOURCE" '[{source:$source,parameters:{city:"Paris",days:3,metric:true}}]' | base64 | tr -d '\n')
QA_PARAM_QUERY=$(jq -rn --arg value "$QA_PARAM_PAYLOAD" '$value|@uri')
control-openhands browser viewport desktop
@@ -137,13 +136,13 @@ control-openhands browser fill 'testid=plugin-0-param-days' 7
control-openhands browser check 'testid=plugin-0-param-metric'
control-openhands assert --field value --equals Tokyo -- browser value 'testid=plugin-0-param-city'
control-openhands assert --field value --equals 7 -- browser value 'testid=plugin-0-param-days'
-control-openhands browser screenshot --feature F19.launch-parameters --name local-delivery-form
control-openhands browser check 'testid=trust-checkbox'
+control-openhands browser screenshot --feature F19.launch-parameters --name local-unsupported
+control-openhands assert --field text --equals 'This local backend does not support plugin parameters. Remove the parameters from the launch link before starting a conversation.' -- browser text 'testid=plugin-parameters-unsupported'
+control-openhands assert --field enabled --equals false -- browser enabled 'testid=start-conversation-button'
control-openhands browser network --clear
-control-openhands browser click 'testid=start-conversation-button'
-control-openhands browser wait-url '/conversations/[0-9a-f-]{36}'
-QA_CREATED_ID=$(control-openhands browser url | jq -er '.url|capture("/conversations/(?<id>[a-f0-9-]{36})").id')
-control-openhands assert --field recent.0.bodyFields.plugins.0 --matches '"parameters":\{"city":"Tokyo","days":7,"metric":true\}' --bug 'recent.0.bodyFields.plugins.0.parameters missing && recent.0.status eq 201' -- browser network --filter '/api/conversations$' --bodies # bug
+control-openhands browser click 'testid=start-conversation-button' --force
+control-openhands assert --field total --equals 0 -- browser network --filter '/api/conversations$'
```
## Gotchas
@@ -154,7 +153,7 @@ control-openhands assert --field recent.0.bodyFields.plugins.0 --matches '"param
- Installing a plugin whose name is already installed points the user to **Update** or **Uninstall**, as Apps do (`F20.install-duplicate`): with the catalog `magic-test` installed, **Add plugin** with `github:OpenHands/extensions` and Path `plugins/magic-test` answers 409 (`browser network --filter 'plugins/install'`) and `browser toasts --history` holds one toast, `This plugin is already installed. Open its card to Update or Uninstall it.`, instead of the server's `force=true` hint; the modal stays open with the typed values (`browser screenshot --feature F19.add-plugin-error --name duplicate-409`). Other install errors still toast the server's reason (`F19.add-plugin-error`). The 409 adds one `http-error` and one `console.error` to `browser errors --app-only`.
- **Uninstall** acts immediately, with no confirmation (MCP server delete and App uninstall confirm).
- `/launch` with a `message` starts the agent as soon as **Start Conversation** is clicked. Prefer links without a message and send `alakazam` yourself; a `city-weather` launch with a message makes the agent call an external weather API (stop it with `testid=stop-button`).
-- Client metadata intentionally strips launch parameters and the Agent Server does not return a live conversation's plugins; neither is delivery proof. Capture the real create request with `control-openhands browser network --bodies --filter '/api/conversations$'` and read `recent.<n>.bodyFields.plugins`: this structured, recursively redacted projection survives the 2000-character display-body cap. It is bounded to 16 KiB and omitted explicitly when too large; default network output omits it. The 2026-10-17 actual Local requests contained coordinates but no parameters (#127 above).
+- Client metadata intentionally strips launch parameters and the Agent Server does not return a live conversation's plugins; neither is delivery proof. Capture the real create request with `control-openhands browser network --bodies --filter '/api/conversations$'` and read `recent.<n>.bodyFields.plugins`: this structured, recursively redacted projection survives the 2000-character display-body cap. It is bounded to 16 KiB and omitted explicitly when too large; default network output omits it. On Local, a launch with parameters sends no create request at all (see `F19.launch-parameters`).
- `conversation start` and launches leave the browser on a conversation page, which has its own `chat-input`; `browser goto /plugins` before the next card step.
- Build deep links with URL-encoded base64 (`%2B`, `%2F`, `%3D`): a raw `+` decodes to a space and gives the invalid-format error. Non-ASCII text must not be used in payloads (`atob` decodes bytes as Latin-1).
- The page loads three queries (marketplace, installed, local) behind skeleton rows; wait for `testid=plugins-toolbar` before counting cards. With no plugins at all the page shows `No plugins found.` (`plugins-empty`), which the bundled catalog makes unreachable; on a Cloud backend **Add plugin** and card actions are disabled (not driven: no Cloud account).
diff --git a/.agents/skills/verify-openhands/references/feature-map/F20-canvas-apps.md b/.agents/skills/verify-openhands/references/feature-map/F20-canvas-apps.md
index 92eba35fb0..233fddf2ce 100644
--- a/.agents/skills/verify-openhands/references/feature-map/F20-canvas-apps.md
+++ b/.agents/skills/verify-openhands/references/feature-map/F20-canvas-apps.md
@@ -64,7 +64,7 @@ Preconditions:
- **Trust confirmation (`F20.enable-confirm`).** Run `control-openhands browser click 'testid=canvas-extension-card-demo-page >> role=switch'` and `control-openhands browser snapshot 'testid=confirmation-modal'`: the paragraph `This app runs trusted JavaScript inside Agent Canvas and can make authenticated requests to the active Agent Server. Review its source and revision before enabling it.`, buttons `Cancel` and `Enable trusted app`. `control-openhands browser click 'testid=confirmation-modal >> testid=cancel-button'` closes it and `aria-checked` stays `false`. Keyboard: `control-openhands browser focus 'testid=canvas-extension-card-demo-page >> role=switch'` and `control-openhands browser press Space` open it again (`browser count 'testid=confirmation-modal'` is `1`); `control-openhands browser press Escape` closes it and `api GET /api/canvas-extensions/installed` still says `"enabled": false`. Backdrop: click the switch again, `control-openhands browser mouse-click 20 500`, then `browser count 'testid=confirmation-modal'` is `0` and the API still says `"enabled": false`. Reopen with the click, then `control-openhands browser click 'testid=confirmation-modal >> testid=confirm-button'`, `control-openhands browser reload` and read `aria-checked`: `true`; `control-openhands browser text 'testid=canvas-extension-card-demo-page'` contains `Enabled`.
- **Rail entry (`F20.rail-entry`, `F20.rail-label`).** After enabling, run `control-openhands browser wait 'testid=sidebar-canvas-extension-demo-page-hello' --timeout 10000`, `control-openhands browser attr 'testid=sidebar-canvas-extension-demo-page-hello' href` (`/extensions/demo-page/hello`) and `control-openhands browser text 'testid=sidebar-canvas-extension-demo-page-hello'`. Expected `Extension demo` (the fixture's `nav_label`); today it reads `Hello from an extension` (see Gotchas).
- **Render the page (`F20.page-render`).** Run `control-openhands browser click 'testid=sidebar-canvas-extension-demo-page-hello' --expect-url '/extensions/demo-page/hello(\?|$)'` and `control-openhands browser snapshot 'role=main[name="Hello from an extension"]'`: heading `Hello from a Canvas Extension` and paragraph `Host API 1 on backend default-local`. Direct URL: `control-openhands browser goto /extensions/demo-page/hello/nested` and the same snapshot show the paragraph `Nested extension path: nested`. `control-openhands browser screenshot --feature F20.page-render --name rail`. The browser tab names the app page: `control-openhands browser eval "document.title"` on `/extensions/demo-page/hello` (and its nested path) should read the page title with an app suffix (`Hello from an extension - OpenHands App`), and leaving the app restores the normal title. Known failure (reproduced 2026-10-17, [#18557](https://github.com/OpenHands/OpenHands/issues/18557)): `document.title` is `OpenHands` after the rail click, on the direct and nested URLs, twice.
-- **App backend view (`F20.app-backend-view`).** Still on `/extensions/demo-page/hello/nested` (where Render the page ends) run `control-openhands browser text 'testid=demo-extension-app-backend-status'`, `control-openhands api GET /server_info --pick capabilities` and `control-openhands api GET /server_info --pick app_backend_ingress_url`. The host hands a page `appBackendView` only when the capabilities list `canvas_app_backend_bridge_v1` and the ingress URL is not `null`: with the pinned server (1.53.0: neither) the line reads `App backend view unavailable`; a server advertising both would make the demo read `App backend view available` (the positive fixture below was driven separately). `control-openhands browser goto /extensions/demo-page/hello` shows the same line, `control-openhands browser snapshot 'role=main[name="Hello from an extension"]'` lists it as the last paragraph, and `control-openhands browser screenshot --feature F20.app-backend-view --name unavailable`. For the positive path use the checked-in [real bridge fixture](../../scripts/fixtures/app-backend-echo/README.md) on an owned Linux amd64 SDK 1.55.1 run. Its helper backs up and arranges only the standard loopback ingress field; restart and doctor the owned run, install/trust the fixture through Apps, then prepare/start the exact backend revision. Open its rail page: a real iframe returns `QA_HTTP_7731` with control-key/cookie presence both false. Require actual negotiated `qa.echo.v1` and echo `QA_ECHO_7731` for the WebSocket branch, not merely an HTTP success. Navigating away disposes the admitted session; a bounded owned backend pause exposes Retry/Open in new tab, and recovery restores the marker. The 2026-10-17 live drive proved these HTTP/session/recovery paths; the three known SDK failures below remain. Stop/uninstall the fixture, restore the saved ingress field and restart/doctor even after a failed drive. The fixture diagnostic text is not a phone-containment specification.
+- **App backend view (`F20.app-backend-view`).** Still on `/extensions/demo-page/hello/nested` (where Render the page ends) run `control-openhands browser text 'testid=demo-extension-app-backend-status'`, `control-openhands api GET /server_info --pick capabilities` and `control-openhands api GET /server_info --pick app_backend_ingress_url`. The host hands a page `appBackendView` only when the capabilities list `canvas_app_backend_bridge_v1` and the ingress URL is not `null`: with the pinned server and no ingress configured (1.55.1: neither) the line reads `App backend view unavailable`; a server advertising both would make the demo read `App backend view available` (the positive fixture below was driven separately). `control-openhands browser goto /extensions/demo-page/hello` shows the same line, `control-openhands browser snapshot 'role=main[name="Hello from an extension"]'` lists it as the last paragraph, and `control-openhands browser screenshot --feature F20.app-backend-view --name unavailable`. For the positive path use the checked-in [real bridge fixture](../../scripts/fixtures/app-backend-echo/README.md) on an owned Linux amd64 SDK 1.55.1 run. Its helper backs up and arranges only the standard loopback ingress field; restart and doctor the owned run, install/trust the fixture through Apps, then prepare/start the exact backend revision. Open its rail page: a real iframe returns `QA_HTTP_7731` with control-key/cookie presence both false. Require actual negotiated `qa.echo.v1` and echo `QA_ECHO_7731` for the WebSocket branch, not merely an HTTP success. Navigating away disposes the admitted session; a bounded owned backend pause exposes Retry/Open in new tab, and recovery restores the marker. The 2026-10-17 live drive proved these HTTP/session/recovery paths; the three known SDK failures below remain. Stop/uninstall the fixture, restore the saved ingress field and restart/doctor even after a failed drive. The fixture diagnostic text is not a phone-containment specification.
- **Active and collapsed rail entry (`F20.rail-active`).** With the demo enabled, run `control-openhands browser goto /apps`; `control-openhands browser attr 'testid=sidebar-canvas-extension-demo-page-hello' aria-current` is `null`; after `control-openhands browser click 'testid=sidebar-canvas-extension-demo-page-hello' --expect-url '/extensions/demo-page/hello(\?|$)'` it is `page`. Run `control-openhands browser click 'testid=sidebar-collapse-toggle'`: `browser visible 'testid=sidebar-canvas-extension-demo-page-hello'` stays `true` (about 40 px wide), `browser attr ... aria-label` and `control-openhands browser tooltip 'testid=sidebar-canvas-extension-demo-page-hello'` read `Hello from an extension`; `control-openhands browser screenshot --feature F20.rail-active --name collapsed` shows the highlighted icon. Expand again with `browser click 'testid=sidebar-collapse-toggle'` (its `aria-label` is `Collapse sidebar` again). Command menu: `browser goto /`, `browser press Control+k`, `browser type 'testid=command-menu >> role=combobox' Hello` and `browser snapshot 'testid=command-menu'` read `No commands found`; `browser press Escape`.
- **Unavailable page (`F20.page-unavailable`).** Run `control-openhands browser goto /extensions/qa-missing/x` and `control-openhands browser snapshot 'role=main'`: heading `Not available` and `This app is disabled, missing, or does not provide this page.` `control-openhands browser goto /extensions/demo-page/nope` shows the same card while the demo is enabled.
- **Phone (`F20.phone`).** Run `control-openhands browser viewport phone`, `control-openhands browser goto /customize` (the URL stays `/customize`), `control-openhands browser click 'testid=extensions-mobile-hub >> testid=sidebar-extensions-/apps' --expect-url '/apps(\?|$)'`, `control-openhands browser bbox 'testid=canvas-extensions-screen'` (`insideViewport` `true`, `pageHorizontalOverflow` `false`) and `control-openhands browser screenshot --feature F20.phone --name apps`. Open the form with `canvas-extensions-add-button`; `control-openhands browser bbox 'testid=add-canvas-extension-modal'` is inside the viewport with no page overflow; close it with `add-canvas-extension-dismiss`. Then `control-openhands browser click 'testid=sidebar-mobile-menu-toggle'`, `control-openhands browser visible 'testid=sidebar-mobile-drawer >> testid=sidebar-canvas-extension-demo-page-hello'` (`true`), `control-openhands browser click 'testid=sidebar-mobile-drawer >> testid=sidebar-canvas-extension-demo-page-hello' --expect-url '/extensions/demo-page/hello(\?|$)'` and `control-openhands browser bbox 'role=main[name="Hello from an extension"]'` (inside, no overflow). Return with `control-openhands browser viewport desktop`.
@@ -98,7 +98,7 @@ Preconditions:
- `control-openhands launch` passes the launcher only a fixed list of environment variables (path, locale, proxy) plus its own flags: `OH_AGENT_SERVER_VERSION=<old>` in the environment is ignored (the run still gets the pinned server); use `--sdk-ref v<version>`. Any server older than 1.47.0 is gated by Canvas before `/apps` renders.
- Stopping the Agent Server makes the next full page load show the **Manage backends** gate; always `restart` and `doctor` afterwards.
- A trailing `browser errors --app-only` sweep shows a `400` per failed install by design; count only other errors.
-- `App backend view unavailable` on the demo page is the expected text while `GET /server_info` lists no `canvas_app_backend_bridge_v1` capability (Agent Server 1.53.0); it does not mean the app failed to load. Cloud backends and the no-backend state never get a backend view either.
+- `App backend view unavailable` on the demo page is the expected text while `GET /server_info` lists no `canvas_app_backend_bridge_v1` capability (Agent Server 1.55.1 with no app-backend ingress configured); it does not mean the app failed to load. Cloud backends and the no-backend state never get a backend view either.
- Known failure (reproduced 2026-10-17, [OpenHands/software-agent-sdk#5648](https://github.com/OpenHands/software-agent-sdk/issues/5648), SDK 1.55.1, `F20.app-backend-view`): a genuine admitted app-backend WebSocket offers `qa.echo.v1`, but the proxy does not negotiate it and Chromium rejects the handshake. Two UI mounts reproduced; a separate direct owned backend control negotiated the protocol and echoed the fixed marker. The positive control is not a proxied UI pass. Platform-matching artifacts and configured ingress are available here; native macOS and other artifact platforms remain separate prerequisites.
- Known failure (reproduced 2026-10-17, [OpenHands/software-agent-sdk#5935](https://github.com/OpenHands/software-agent-sdk/issues/5935), SDK 1.55.1, `F20.app-backend-view`): with an owned JSON `app_backend_public_url` plus explicit `OH_APP_BACKEND_PUBLIC_URL_IS_NONE=1`, the actual server still advertises the ingress/capability and admits the app view. Expected: the null override suppresses it. The conflicting standard configuration sources were arranged and restored, not inferred from a default server.
-- Known failure (reproduced 2026-10-17, [OpenHands/software-agent-sdk#5914](https://github.com/OpenHands/software-agent-sdk/issues/5914), SDK 1.55.1, `F20.app-backend-view`): the real admitted HTTP response joins duplicate Date values and exposes `Server: uvicorn, QAOwnedBridge/1 Python/3.13.5`. Both original and checked-in fixture drives reproduced; a direct request alone was not counted as proxy proof. Preserve the expected valid single response headers; no paired fix retirement is claimed.
+- Known failure (reproduced twice 2026-10-19, [OpenHands/software-agent-sdk#5914](https://github.com/OpenHands/software-agent-sdk/issues/5914), SDK 1.55.1, `F20.app-backend-view`): the real admitted HTTP response joins duplicate Date values and exposes `Server: uvicorn, QAOwnedBridge/1 Python/<version>` (the fixture result's `http.date` and `http.server`). The SDK fix ([software-agent-sdk#5916](https://github.com/OpenHands/software-agent-sdk/pull/5916)) is not in the pinned 1.55.1; re-drive when the pin moves past it. Checked-in fixture drives reproduced; a direct request alone was not counted as proxy proof. Preserve the expected valid single response headers; no paired fix retirement is claimed.
diff --git a/.agents/skills/verify-openhands/references/feature-map/F21-automations-dashboard.md b/.agents/skills/verify-openhands/references/feature-map/F21-automations-dashboard.md
index 57a517637b..bc14cf2880 100644
--- a/.agents/skills/verify-openhands/references/feature-map/F21-automations-dashboard.md
+++ b/.agents/skills/verify-openhands/references/feature-map/F21-automations-dashboard.md
@@ -46,7 +46,7 @@ Source: `src/routes/automations-list.tsx`, `src/components/features/automations/
- `F21.run-now`: Play or kebab **Run now** on an enabled automation dispatches a run with the toast **Automation dispatched**; the card moves through Pending/Running to the result. Both are disabled on a disabled automation.
- `F21.run-now-error`: a failed dispatch shows exactly one error toast with the API message; when the automation is gone (`Automation not found`), its card then drops out of the dashboard without a reload.
- `F21.toggle`: kebab **Turn on** / **Turn off** flips the enabled flag immediately (no confirmation on the dashboard) and moves the automation between the Active and Inactive groups.
-- `F21.edit`: kebab **Edit** opens the setup page in edit mode (`/automations/setup?automationId=<id>`) prefilled from the automation, with **Test**, **Save** and **Close** in its toolbar; Save sends one PATCH and toasts **Automation updated**, Close returns to `/automations` (field-level editing is F23). For a script (tarball) automation it should summarize the saved bundle (**Existing code bundle**). Known failure (reproduced 2026-10-13, #18341): from the kebab it shows generated starter Python and a default setup script instead; a cold load of the same URL shows the bundle summary.
+- `F21.edit`: kebab **Edit** opens the setup page in edit mode (`/automations/setup?automationId=<id>`) prefilled from the automation, with **Test**, **Save** and **Close** in its toolbar; Save sends one PATCH and toasts **Automation updated**, Close returns to `/automations` (field-level editing is F23). For a script (tarball) automation it summarizes the saved bundle (**Existing code bundle** with the upload id) instead of a code editor, from the kebab and on a cold load alike, and saving keeps that bundle (F23.edit-save).
- `F21.export`: kebab **Export** downloads `<name-slug>.automation.json` with `version`, `kind: "automation"` and the spec.
- `F21.delete`: kebab **Delete** asks `Delete "<name>"? This cannot be undone.`; Cancel and the X keep it, Delete removes it for good, and Escape dismisses the dialog.
- `F21.saved-drafts`: saved setup drafts (#17242) appear in a **Drafts** group above Active/Inactive, one row each with the name (Resume), a pencil (**Edit**), Play (**Test draft**, tooltip **Test Run**; disabled for event-triggered or non-dispatchable drafts) and a trash (**Delete draft**). Resume and Edit open `/automations/setup?draftId=<id>`; Test dispatches a test run (toast **Test run started.**) on a materialized DRAFT-state automation that the groups do not list; Delete asks **Delete draft?** (`Delete “<name>”? This removes the saved setup draft, but does not delete any final automations.`) with Cancel / **Delete draft** and toasts **Draft deleted.**, also removing the draft's materialized test automation. The confirmation is the named modal dialog **Delete draft?** (the automation delete dialog with draft copy): it opens with focus on Cancel, keeps Tab inside, closes on Escape, X, Cancel or a backdrop click and returns focus to the trash button.
@@ -102,7 +102,7 @@ Preconditions:
- **Filtered empty and Clear filters (`F21.filtered-empty`).** With `zzqa-nomatch` in the search box run `control-openhands browser text 'testid=automations-filtered-empty'` (`No automations match these filters` / `Clear filters`), `control-openhands browser click 'testid=automations-clear-filters'`, `control-openhands browser value 'role=textbox[name="Search automations..."]'` (empty) and the row count (`2`).
- **Filters (`F21.filters`).** Run `control-openhands browser click 'testid=automations-filters >> testid=dropdown-trigger >> nth=0'`, `control-openhands browser snapshot 'testid=automations-filters-menu'` (Filter by status `All statuses`, Filter by trigger `All triggers`, Sort automations `Latest run`), `control-openhands browser click 'testid=automations-filter-status >> testid=dropdown-trigger'`, `control-openhands browser click 'testid=automations-filter-status-disabled'`, then `control-openhands browser text 'testid=automations-filters >> testid=dropdown-trigger >> nth=0'` (`Filters 1`) and the row text (only QA_F21 Beta). The popover stays open: `control-openhands browser count 'testid=automations-filters-menu'` is `1` and `control-openhands browser attr 'testid=automations-filters >> testid=dropdown-trigger >> nth=0' aria-expanded` is `true`. In the same popover run `control-openhands browser click 'testid=automations-filter-trigger >> testid=dropdown-trigger'` and `control-openhands browser click 'testid=automations-filter-trigger-schedule'`: the badge is `Filters 2`, `testid=automations-filtered-empty` appears and the menu count is still `1` (`control-openhands browser screenshot --feature F21.filters --name two-filters`). `control-openhands browser click 'testid=automations-filters-reset'`: both rows return, the button reads `Filters` and the popover stays open. Trigger **Event-driven** (`automations-filter-trigger-event`, chosen in the same open popover) alone leaves only QA_F21 Beta. An outside click closes the popover: `control-openhands browser click 'role=heading[name="Dashboard"]'`, then the menu count is `0` and `aria-expanded` is `false`. Run `control-openhands browser reload` to clear the filter (filters are page state): the next bullets need Alpha visible.
- **Turn off (`F21.toggle`).** In list view run `control-openhands browser click '[data-testid^="automation-list-row-"] >> has-text=QA_F21 Alpha >> role=button[name="Automation actions"]'`, `control-openhands browser click 'role=list >> has-text=Export >> role=button[name="Turn off"]'`, `control-openhands browser count 'testid=confirmation-modal'` (`0`, no confirmation), `control-openhands browser reload`. The page shows **Inactive 2** with the tile at `0 active`, and Play reports `enabled` `false`.
-- **Edit from the dashboard (`F21.edit`).** Turn off (above) left both automations in the Inactive group, as Sort needs. Open Alpha's kebab and `control-openhands browser click 'role=list >> has-text=Export >> role=button[name="Edit"]' --expect-url '/automations/setup'`: the URL is `/automations/setup?automationId=<id>`, `control-openhands browser value 'testid=automation-setup-name'` is `QA_F21 Alpha`, `automation-setup-prompt` holds the prompt, `automation-setup-custom-schedule` is `0 0 1 1 *` (Frequency Custom, timezone UTC), Timeout `600`, and `browser testids 'testid=automation-setup-toolbar'` lists `automation-setup-test`, `automation-setup-save` and `automation-setup-close`; `control-openhands conversation list` is unchanged. Run `control-openhands browser fill 'testid=automation-setup-name' 'QA_F21 Zulu'`, `control-openhands browser network --clear`, `control-openhands browser click 'testid=automation-setup-save'` and `control-openhands browser toasts --history` (`Automation updated`); `control-openhands browser network --bodies --filter 'automation/v1'` shows one `PATCH /api/automation/v1/<id>` with `name`, `trigger`, `model: null` (Active profile kept), `agent_profile_id: null`, `timeout` and `prompt`, and `control-openhands api GET /api/automation/v1` shows `QA_F21 Zulu` with the schedule unchanged. The toast sits over the toolbar and stays while the pointer rests on it (see Gotchas): run `control-openhands browser hover 'testid=automation-setup-name'` and wait about 5 s, then `control-openhands browser click 'testid=automation-setup-close' --expect-url '/automations$'`. At phone width the same kebab Edit on QA_F21 Beta (event trigger) shows Event selected with Event source GitHub and no horizontal overflow (`browser bbox 'testid=automation-setup-form'`). Script automation: Known failure (reproduced 2026-10-13, #18341): for a tarball automation (create one with F22's Upload tarball bullet), kebab Edit shows `Custom Python` with generated starter code (`browser value 'testid=automation-setup-custom-code'` contains `def fire_callback` and an empty prompt) and `browser count 'testid=automation-setup-existing-bundle'` is `0`; expected the **Existing code bundle** summary, which a cold `browser goto '/automations/setup?automationId=<id>'` does show.
+- **Edit from the dashboard (`F21.edit`).** Turn off (above) left both automations in the Inactive group, as Sort needs. Open Alpha's kebab and `control-openhands browser click 'role=list >> has-text=Export >> role=button[name="Edit"]' --expect-url '/automations/setup'`: the URL is `/automations/setup?automationId=<id>`, `control-openhands browser value 'testid=automation-setup-name'` is `QA_F21 Alpha`, `automation-setup-prompt` holds the prompt, `automation-setup-custom-schedule` is `0 0 1 1 *` (Frequency Custom, timezone UTC), Timeout `600`, and `browser testids 'testid=automation-setup-toolbar'` lists `automation-setup-test`, `automation-setup-save` and `automation-setup-close`; `control-openhands conversation list` is unchanged. Run `control-openhands browser fill 'testid=automation-setup-name' 'QA_F21 Zulu'`, `control-openhands browser network --clear`, `control-openhands browser click 'testid=automation-setup-save'` and `control-openhands browser toasts --history` (`Automation updated`); `control-openhands browser network --bodies --filter 'automation/v1'` shows one `PATCH /api/automation/v1/<id>` with `name`, `trigger`, `model: null` (Active profile kept), `agent_profile_id: null`, `timeout` and `prompt`, and `control-openhands api GET /api/automation/v1` shows `QA_F21 Zulu` with the schedule unchanged. The toast sits over the toolbar and stays while the pointer rests on it (see Gotchas): run `control-openhands browser hover 'testid=automation-setup-name'` and wait about 5 s, then `control-openhands browser click 'testid=automation-setup-close' --expect-url '/automations$'`. At phone width the same kebab Edit on QA_F21 Beta (event trigger) shows Event selected with Event source GitHub and no horizontal overflow (`browser bbox 'testid=automation-setup-form'`). Script automation: for a tarball automation (create one with F22's Upload tarball bullet), kebab Edit shows `Custom Python` with the **Existing code bundle** summary: `browser count 'testid=automation-setup-existing-bundle'` is `1`, `browser text 'testid=automation-setup-existing-bundle-name'` is the upload id (its `title` is the full `oh-internal://uploads/<id>` path) and `browser count 'testid=automation-setup-custom-code'` is `0`; a cold `browser goto '/automations/setup?automationId=<id>'` shows the same summary.
- **Sort (`F21.sort`).** With QA_F21 Zulu (1 run) and QA_F21 Beta (no runs) both inactive, open Filters, `control-openhands browser click 'testid=automations-sort >> testid=dropdown-trigger'`, `control-openhands browser click 'testid=automations-sort-name'`, then `control-openhands browser text '[data-testid^="automation-list-row-"] >> nth=0'`. Beta comes first, the badge reads `Filters 1` and the popover stays open. Fill the search with `zzqa-nomatch` and click `testid=automations-clear-filters` (a click outside the popover, which closes it): the badge stays `Filters 1` and Beta stays first (the sort is kept). Open Filters and the sort field again and choose `automations-sort-runs`; Zulu comes first. Close the popover with `control-openhands browser click 'role=heading[name="Dashboard"]'` (`browser count 'testid=automations-filters-menu'` is `0`).
- **Export (`F21.export`).** Open Zulu's kebab, `control-openhands browser click 'role=list >> has-text=Export >> role=button[name="Export"]'`, then `control-openhands browser downloads --last 1 --inspect`. The download path ends in `qa-f21-zulu.automation.json` (under `<run>/private/downloads/`); the inspected head holds `"version": 1`, `"kind": "automation"` and a `spec` with name, trigger, `enabled`, prompt, `model` (`null` for Active profile), timeout and timezone.
- **Open the detail page (`F21.open-detail`).** Kebab `role=list >> has-text=Export >> role=button[name="View"]`, `control-openhands browser click '[data-testid^="automation-list-row-"] >> has-text=QA_F21 Zulu >> role=link >> nth=0'`, `control-openhands browser click '[data-testid^="automation-card-"] >> has-text=QA_F21 Zulu >> role=heading'` (grid: switch first with `control-openhands browser click 'testid=automations-view-toggle'` and `control-openhands browser click 'testid=automations-view-toggle-grid'`; the later bullets use cards), and `control-openhands browser focus '[data-testid^="automation-card-"] >> has-text=QA_F21 Beta'` + `control-openhands browser press Enter` each make `control-openhands browser url` read `/automations/<that automation's id>`. Go back with `control-openhands browser back`.
diff --git a/.agents/skills/verify-openhands/references/feature-map/F23-automation-detail.md b/.agents/skills/verify-openhands/references/feature-map/F23-automation-detail.md
index 4d40b8ccf5..0a29330d3c 100644
--- a/.agents/skills/verify-openhands/references/feature-map/F23-automation-detail.md
+++ b/.agents/skills/verify-openhands/references/feature-map/F23-automation-detail.md
@@ -47,10 +47,10 @@ Source: `src/routes/automation-detail.tsx`, `src/components/features/automations
- `F23.export`: kebab **Export** downloads `<name-slug>.automation.json` (`version`, `kind: "automation"`, `spec`); an automation's first repository and its plugins go into `spec` as `repository`, `branch` and `plugins`.
- `F23.tarball`: kebab **Tarball** downloads the automation's code bundle.
- `F23.delete`: kebab **Delete** asks `Delete "<name>"? This cannot be undone.`; Cancel keeps it, Delete removes it and returns to `/automations`.
-- `F23.edit-open`: kebab **Edit** opens the setup page `/automations/setup?automationId=<id>` (#17688; the **Edit automation** modal is gone) prefilled with Name, Prompt, agent profile and LLM profile pickers, the trigger (Schedule with Frequency, or Event), timezone and Timeout seconds, with **Test**, **Save** and **Close** in the toolbar and a docked composer; no conversation exists until a prompt is sent.
+- `F23.edit-open`: kebab **Edit** opens the setup page `/automations/setup?automationId=<id>` (#17688; the **Edit automation** modal is gone) prefilled with Name, Prompt, agent profile and LLM profile pickers, the trigger (Schedule with Frequency, or Event), timezone and Timeout seconds, with **Test**, **Save** and **Close** in the toolbar and a docked composer; no conversation exists until a prompt is sent. A script (tarball or bundle) automation shows `Custom Python` with the **Existing code bundle** summary (`automation-setup-existing-bundle`, the upload id) instead of a code editor.
- `F23.edit-close`: **Close** leaves without saving and lands on `/automations` (the dashboard, not the detail page); Escape does nothing on the page.
- `F23.edit-validation`: an empty prompt is refused with the toast **Add a prompt before continuing.** (no request); a bad cron, a cron that never runs or a timeout above 1800 is sent to the API, refused with `422`, and its message is the single error toast (`Value error, Invalid cron expression: not a cron`, `Value error, Cron expression cannot produce any future fire times: 0 0 31 2 *`, `Value error, timeout must not exceed 1800 seconds (30 minutes)`); nothing is saved.
-- `F23.edit-save`: Save sends one PATCH (also when nothing changed), shows **Saving…** then **Saved just now**, the toast **Automation updated**, and the detail page and API reflect it; an Active profile model stays `null`.
+- `F23.edit-save`: Save sends one PATCH (also when nothing changed), shows **Saving…** then **Saved just now**, the toast **Automation updated**, and the detail page and API reflect it; an Active profile model stays `null`. On a script (tarball or bundle) automation, including one without a `setup_script_path`, Save keeps the existing bundle: it asks for no setup script path, setup script or code, the PATCH carries no bundle fields, and `tarball_path`, `entrypoint` and `setup_script_path` read back unchanged; the summary stays after each edit and a reload.
- `F23.edit-test`: on the edit page **Test** dispatches the automation as last saved (`POST /api/automation/v1/<id>/dispatch`, no PATCH first) and shows `Test run started.` on the form; unsaved edits are not part of that run. Whether it should test the current form is a product decision (Gotchas).
- `F23.edit-dirty-state`: the edit page shows no save state for unsaved edits (no **Unsaved changes**, which create mode shows, F28); the label appears only while saving (**Saving…**) and after (**Saved just now** or **Save failed**). **Close** discards the edits.
- `F23.edit-schedule`: Frequency offers Once, Hourly, Daily, Weekdays (Mon–Fri), Weekly and Custom; Weekly adds **Day of week** and **At**; they are saved as a cron expression and read back after a reload. Switching to Custom starts the Custom field from the cron of the preset it replaces (Weekly, Wednesday at `10:30` → `30 10 * * 3`), and a saved cron that matches a preset reads back as that preset, not as Custom.
@@ -58,7 +58,7 @@ Source: `src/routes/automation-detail.tsx`, `src/components/features/automations
- `F23.edit-cron`: a custom cron schedule shows Frequency **Custom** with an editable **Custom** cron field and a timezone select.
- `F23.edit-event-readonly`: for an event automation the edit page shows Event selected with Event source, event key and filter controls (editable now, no longer read-only text) and no schedule fields.
- `F23.edit-save-error`: a failed save shows exactly one error toast with the API message (`Automation not found` for an automation deleted meanwhile), the save state reads **Save failed** and the edits stay on the page; the toolbar **Test** on that page also toasts `Automation not found`.
-- `F23.phone`: at 390×844 the detail page has no horizontal overflow, and the Edit page fits with its title, Test, Save and Close reachable. Known failure (reproduced 2026-10-17, [#18613](https://github.com/OpenHands/OpenHands/issues/18613)): at 390×844, the 21-character name `QA_Inert_Event_Detail` clips the header switch/actions in both LTR (switch x385.609–425.609) and RTL (x−30.797–9.203). Document width remains 390, so scrolling cannot reveal them; the short name `QA` fits. Space still toggles the switch, which does not establish pointer accessibility.
+- `F23.phone`: at 390×844 the detail page has no horizontal overflow, and the Edit page fits with its title, Test, Save and Close reachable. Below 640 px the header stacks: the name (wrapping when long) and badge on one row, then **Run now**, the switch and the kebab, all inside the viewport in LTR and RTL; a long name such as `QA_Inert_Event_Detail` keeps the switch and kebab reachable by pointer (LTR switch x≈136–176, kebab x≈184–216; document width 390). From 640 px up the name and actions share one row, actions on the right.
## How to get to it (user POV)
@@ -104,10 +104,11 @@ Preconditions:
- **Task outcome badge (`F23.run-task-outcome`).** Arrange (API, arrange only) `control-openhands api POST /api/automation/v1/preset/prompt --write --data '{"name":"QA_F23 Blocked","prompt":"Do not run any tools. Call the finish tool right away with status \"blocked\" and outcome_summary \"QA blocked on purpose\".","trigger":{"type":"cron","schedule":"0 0 1 1 *","timezone":"UTC"},"enabled":true}'` (a good thing to do while the Fail run is still running). Open QA_F23 Blocked, click `role=button[name="Run now"]`, and poll `control-openhands browser text 'testid=automation-activity-log'` until it leaves Running (about 35 s): `<time>`, `QA blocked on purpose`, a cost such as `$0.0016`, `Blocked`; `control-openhands browser testids 'testid=automation-activity-log'` lists `run-status-icon-warning` (not `-completed`) although `api GET /api/automation/v1/<id>/runs` has `status` `COMPLETED` and `run_metadata.finish_tool_response.status` `blocked`. `control-openhands browser click 'testid=automation-activity-log >> title=View logs'` and `browser snapshot 'role=dialog'`: `Run` `COMPLETED`, `Task` `Blocked` / `QA blocked on purpose` / `Task metadata` `{ "message": ... }`. Screenshot with `--feature F23.run-task-outcome --name blocked`, Escape. `partial_success` (**Partial**, `run-status-icon-warning`) and `unknown` (**Needs review**, `run-status-icon-needs-review`) follow the same path with the same prompt shape. The normal automation 1.19.0 prompt/plugin presets configure SDK 1.53.0 FinishTool with the required `TaskOutcome.status`, limited to `success`, `partial_success`, `blocked`, `failed` and `unknown`; SDK custom response schemas are a separate producer boundary. Its normal prompt path therefore does not establish the new fallback for a finish response without a string status. That positive boundary needs a genuine completed run from a legacy or custom-schema producer whose real finish arguments have that shape; a run with no finish metadata checks only the ordinary Completed fallback. Arbitrary unrecognised strings and summary retention on a failed run need the corresponding genuine producer too. Until available, record those checks as blocked and do not treat injected run metadata or mocked LLM output as live proof (#17682).
- **Event automation (`F23.prompt-collapse`, `F23.configuration-event`, `F23.plugins-repo`, `F23.export`).** Open QA_F23 Event. `control-openhands browser bbox 'testid=automation-prompt-content'` is 240 px high, `testid=automation-prompt-fade` is present and `control-openhands browser text 'testid=automation-prompt-toggle'` is `View More`; screenshot with `--feature F23.prompt-collapse --name collapsed`. Click the toggle: height 360, no fade, text `View Less`; click again to collapse. `control-openhands browser text 'role=heading[name="Configuration"] >> xpath=ancestor::div[2]'` reads `Repositories qa-example/qa-repo main`, `Trigger Event`, `Event Source github`, `Event Type pull_request.opened`, `Filter pull_request.draft == `false` && pull_request.base.ref == re…`, `Show more`, `LLM profile Active profile`. `control-openhands browser click 'role=button[name="Show more"]'` shows the full filter and `control-openhands browser attr 'role=button[name="Show less"]' aria-expanded` is `true`. `control-openhands browser text 'testid=automation-repository'` is `qa-example/qa-repo` / `main` (the fixture's `https://github.com/qa-example/qa-repo` as `owner/repo`, with its branch badge) and `control-openhands browser text 'role=heading[name="Plugins"] >> xpath=ancestor::div[2]'` is `Plugins` / `github:qa-example/qa-plugin`; screenshot with `--feature F23.plugins-repo --name event-repo-plugins`. Export keeps them: `control-openhands browser click 'role=button[name="Automation actions"]'`, `control-openhands browser click 'role=list >> has-text=Tarball >> role=button[name="Export"]'` and `control-openhands browser downloads --last 1 --contains '"branch": "main"'` (`qa-f23-event.automation.json`, `contains` `true`); `--contains '"repository": "https://github.com/qa-example/qa-repo"'` and `--contains 'github:qa-example/qa-plugin'` are `true` too (the `--inspect` head ends inside the long prompt).
- **Event trigger in Edit (`F23.edit-event-readonly`).** On QA_F23 Event open Edit: Event is selected, `browser testids 'testid=automation-setup-form' --filter event` lists `automation-setup-event-source` (GitHub), `automation-setup-event-source-toggle`, `automation-setup-event-public-url-notice`, `automation-setup-event-key`, `automation-setup-event-filter` and `automation-setup-event-filter-remove` (the fixture has a filter), and there are no Frequency controls for a schedule. The trigger is editable here (it was read-only text in the old modal). Screenshot with `--feature F23.edit-event-readonly --name event`, then Close.
-- **Phone (`F23.phone`).** Run `control-openhands browser viewport phone`, open QA_F23 Pong, `control-openhands browser bbox 'role=heading[name="QA_F23 Pong"] >> xpath=ancestor::div[3]'` (`insideViewport` `true`, `pageHorizontalOverflow` `false`) and `control-openhands browser screenshot --feature F23.phone --name detail` (the name and Run now wrap). Open Edit (kebab, `--expect-url '/automations/setup'`), then `control-openhands browser bbox 'testid=automation-setup-save'` and `... 'testid=automation-setup-close'` (both `insideViewport` `true`, Save at `y` about 65) and `control-openhands browser bbox 'testid=automation-setup-form'` (`pageHorizontalOverflow` `false`); screenshot with `--feature F23.phone --name edit-setup-page`. Close and `control-openhands browser viewport desktop`.
+- **Phone (`F23.phone`).** Run `control-openhands browser viewport phone`, open QA_F23 Pong, `control-openhands browser bbox 'role=heading[name="QA_F23 Pong"] >> xpath=ancestor::div[3]'` (`insideViewport` `true`, `pageHorizontalOverflow` `false`) and `control-openhands browser screenshot --feature F23.phone --name detail` (the name and Run now wrap). Long name, LTR: arrange `QA_Inert_Event_Detail` as in the header check below (English), `control-openhands browser goto /automations/<id>` and `control-openhands browser eval "(()=>{const r=document.querySelector('[role=switch]').getBoundingClientRect();const k=document.querySelector('button[aria-label=\"Automation actions\"]').getBoundingClientRect();return {sw:[r.left,r.right],kebab:[k.left,k.right],scrollWidth:document.documentElement.scrollWidth}})()"`: both inside `0`–`390`, `scrollWidth` `390`; clicking `role=switch[name="Turn on"]` turns it on (and back off); then `api DELETE /api/automation/v1/<id> --write`. The header check below repeats this in Arabic (RTL). Open Edit on Pong (kebab, `--expect-url '/automations/setup'`), then `control-openhands browser bbox 'testid=automation-setup-save'` and `... 'testid=automation-setup-close'` (both `insideViewport` `true`, Save at `y` about 65) and `control-openhands browser bbox 'testid=automation-setup-form'` (`pageHorizontalOverflow` `false`); screenshot with `--feature F23.phone --name edit-setup-page`. Close and `control-openhands browser viewport desktop`.
- **Run now and save errors (`F23.run-now-error`, `F23.edit-save-error`).** Arrange a stale page: open QA_F23 Fail, then `control-openhands api DELETE /api/automation/v1/<Fail id> --write`. Run `control-openhands browser toasts --history --clear` (empties the toast history), click `role=button[name="Run now"]`, then `control-openhands browser toasts --history` and `control-openhands browser screenshot --feature F23.run-now-error --name toasts`: exactly one toast, `Automation not found`, and no `Request failed with status code 404`. Expected: the page then shows **Automation not found**: `control-openhands browser wait 'role=heading[name="QA_F23 Fail"]' --state detached --timeout 15000` succeeds (the not-found state takes about 7 s, as in States) and `control-openhands browser snapshot 'role=paragraph >> text=Automation not found >> xpath=..'` lists link `Back to Automations` (do not use `wait-text 'Automation not found'` here: the toast has the same text). Once the heading is gone, run `control-openhands browser toasts --history` again: it still holds only `Automation not found` (a late second toast, `Request failed with status code 404`, would come from the page's own refetch of the deleted automation). Known failure: the wait times out and the page keeps heading `QA_F23 Fail`, `Active` and an enabled **Run now** (`control-openhands browser enabled 'role=button[name="Run now"]'` is `true`) until a reload ([#18062](https://github.com/OpenHands/OpenHands/issues/18062)). The save error does not reuse that page (it is gone once [#18062](https://github.com/OpenHands/OpenHands/issues/18062) is fixed): arrange a second fixture, `control-openhands api POST /api/automation/v1/preset/prompt --write --data '{"name":"QA_F23 Stale","prompt":"Reply with the single word: pong.","trigger":{"type":"cron","schedule":"0 0 1 1 *","timezone":"UTC"},"enabled":false}'` (`body.id` is `<stale-id>`), open QA_F23 Stale, its kebab and `role=list >> has-text=Tarball >> role=button[name="Edit"]` (`--expect-url '/automations/setup'`). With the edit page open, `control-openhands api DELETE /api/automation/v1/<stale-id> --write` and `control-openhands browser toasts --history --clear`. Fill `testid=automation-setup-name` with `QA_F23 Stale renamed` and click `testid=automation-setup-save`: the page keeps the edit (`browser value 'testid=automation-setup-name'` still `QA_F23 Stale renamed`) and `browser toasts --history` holds exactly one toast, `Automation not found`. Screenshot with `--feature F23.edit-save-error --name toast`. Move the pointer off the toast (`browser hover 'testid=automation-setup-name'`, about 5 s), `browser toasts --history --clear` and click `testid=automation-setup-test`: again exactly one toast, `Automation not found`. Close. `browser errors --app-only` lists the induced `POST /api/automation/v1/<Fail id>/dispatch`, `PATCH /api/automation/v1/<stale-id>` and `POST /api/automation/v1/<stale-id>/dispatch` 404s; both fixtures are already deleted.
- **Load more runs (`F23.activity-log-more`).** Not run: it needs more than 20 runs of one automation (about 35 s and one model call each). With them, `control-openhands browser click 'testid=automation-activity-log >> role=button[name="Load more"]'` grows the list from 20 to 40 rows.
- **Script automation (`F23.script-section`).** Arrange a script automation through F22's bundle entry (UI arrange; F22 owns the creation proof): `control-openhands browser goto /automations/templates`, `control-openhands browser click 'testid=recommended-automation-card-news-digest' --expect-url '/automations/new/news-digest'`, `control-openhands browser fill 'testid=setup-field-schedule' '0 0 1 1 *'`, `control-openhands browser click 'testid=setup-continue-button'`, `control-openhands browser wait 'testid=setup-review' --timeout 10000`, `control-openhands browser click 'testid=setup-continue-button' --expect-url '/automations/[0-9a-f-]{8,}' --timeout 60000`. The detail page of **Daily news digest** shows `role=heading[name="Script"]` and no Prompt (`browser count 'role=heading[name="Prompt"]'` is `0`); `control-openhands browser text 'testid=automation-script-entrypoint'` is `python3 main.py`, and `browser eval "Array.from(document.querySelectorAll('[data-testid=automation-script-file]')).map(e=>e.innerText.split('\n').slice(0,2).join(' | '))"` lists `main.py | Entrypoint` first, then `config.json` (both with line-numbered source). Screenshot with `--feature F23.script-section --name news-digest`. `automation-script-unavailable` (bundle unreadable) was not reached. Do not Run now: the script fetches public feeds and starts a model-backed conversation. Delete it in the Delete bullet.
+- **Script automation Edit and Save (`F23.edit-open`, `F23.edit-save`).** On Daily news digest (`control-openhands api GET /api/automation/v1/<id>` shows a `tarball_path` and `setup_script_path` `null`) open the kebab and `role=list >> has-text=Tarball >> role=button[name="Edit"]` (`--expect-url '/automations/setup'`): `browser count 'testid=automation-setup-existing-bundle'` is `1` and `browser count 'testid=automation-setup-custom-code'` is `0`. Fill `testid=automation-setup-custom-schedule` with `0 0 2 1 *` (a preset such as `testid=automation-setup-frequency-weekdays` works the same), `browser toasts --history --clear`, `browser network --clear`, click `testid=automation-setup-save` and screenshot with `--feature F23.edit-save --name script-bundle`: `browser toasts --history` holds only `Automation updated` (no `Add a setup script path before continuing.`), `browser network --bodies --filter 'automation/v1'` shows one `PATCH … 200` with `name`, `trigger`, `model`, `agent_profile_id` and `timeout` but no bundle fields, and the `api GET` keeps the same `tarball_path`, `entrypoint` and `setup_script_path` `null` with the new schedule. A second edit (rename) saves the same way, and after `browser reload` the summary is still shown. F22's Upload tarball automation (`setup_script_path` `null`) behaves the same, from the dashboard kebab too and at 390×844. Close.
- **Edit-mode Test and unsaved edits (`F23.edit-test`, `F23.edit-dirty-state`).** Open QA_F23 Event and its kebab **Edit** (`--expect-url 'automationId='`). Fill `testid=automation-setup-name` with `QA_F23 Unsaved`: `control-openhands browser count 'testid=automation-setup-save-state'` is `0`. `control-openhands browser network --clear`, click `testid=automation-setup-test`; `control-openhands browser network --filter 'automation/v1'` shows `POST …/<id>/dispatch 201` and no `PATCH`, and the form shows `Test run started.` Move the pointer off the toast (`control-openhands browser hover 'testid=automation-setup-prompt'`) and click `testid=automation-setup-close` (`--expect-url '/automations$'`): the card still reads `QA_F23 Event`. The dispatched run is not checked here; Delete below removes it with the automation.
- **Delete (`F23.delete`).** Open QA_F23 Event, click the kebab and `role=list >> has-text=Tarball >> role=button[name="Delete"]`. `control-openhands browser snapshot 'role=dialog[name="Delete automation"]'` shows `Delete "QA_F23 Event"? This cannot be undone.`, `Cancel` and `Delete` (the same dialog as F21.delete). `control-openhands browser click 'role=dialog[name="Delete automation"] >> role=button[name="Cancel"]'` keeps the page (`control-openhands browser count 'role=dialog[name="Delete automation"]'` is `0`). Reopen and `control-openhands browser click 'role=dialog[name="Delete automation"] >> role=button[name="Delete"]' --expect-url '/automations(\?|$)'`; `control-openhands api GET /api/automation/v1` no longer lists it. Delete QA_F23 Pong, QA_F23 Blocked and Daily news digest the same way (and QA_F23 Fail if the error bullet was skipped); `api GET /api/automation/v1` shows `"total": 0`.
- **Errors.** After each group `control-openhands browser errors --app-only` shows no page errors; the only app errors are the induced ones (404s for the unknown id and the deleted fixture, 502s while the service is stopped).
@@ -117,9 +118,9 @@ Preconditions:
- **Custom nested setup (`F23.script-section`, `F23.run-logs-modal`; creation in F22).** Arrange a reviewed tarball containing `scripts/setup.sh` that writes `.qa-setup-marker` and `main.py` that asserts that marker exists. Upload with `api POST /api/automation/v1/uploads?name=qa-nested-setup --write --data-file <tarball> --content-type application/gzip`; create with `entrypoint: python3 main.py`, `setup_script_path: scripts/setup.sh`, active state and yearly cron. Open the detail page and click **Run now**; correlate its `bash_command_id` with `api GET '/api/bash/bash_events/search?command_id__eq=<id>&limit=100'`. Expected: the configured nested setup runs before the entrypoint. Known failure, reproduced 2026-10-12 against automation 1.19.3: the real BashOutput exits 1 with the marker assertion, proving the setup step was skipped ([OpenHands/automation#343](https://github.com/OpenHands/automation/issues/343)). Only an owned marker is written; no external service or model is used.
- **Selected profile on a prompt preset (`F23.configuration`, `F23.run-linked-conversation`).** Arrange a bare saved Agent Profile (`tools: []`, unique persona) referencing an offline LLM profile whose base URL is the closed loopback `http://127.0.0.1:9`; set zero retries and disable title generation. Create an owned active annual-cron prompt preset, PATCH its `agent_profile_id` to the bare profile, then click **Run now** in Canvas. Inspect the linked conversation's `launched_agent_profile`, tool names and persona without revealing secrets. Expected: the chosen profile governs that conversation. Known failure, reproduced 2026-10-12 against automation 1.19.3/Agent Server 1.54.0: provenance is null, the selected persona is absent, and default terminal/file-editor/task-tracker/finish tools replace the bare selection ([OpenHands/automation#557](https://github.com/OpenHands/automation/issues/557)). The deliberate loopback connection failure is incidental to this metadata check; no provider is reachable.
-Header viewport regression (`F23.phone`):
+Header viewport check (`F23.phone`, RTL with a long name):
-```sh known-bug F23.phone detail-header-clipping
+```sh
QA_HEADER_ID=$(control-openhands api POST /api/automation/v1/preset/prompt --write --data '{"name":"QA_Inert_Event_Detail","prompt":"Reply pong. Do not run any tools.","trigger":{"type":"event","source":"qa-never-delivered","on":"qa.never"},"enabled":false}' | jq -er '.body.id')
trap 'control-openhands api DELETE "/api/automation/v1/$QA_HEADER_ID" --write >/dev/null; control-openhands browser goto /settings/app >/dev/null; control-openhands browser choose testid=language-input English >/dev/null; control-openhands browser click testid=submit-button >/dev/null; control-openhands browser viewport desktop >/dev/null' EXIT
control-openhands browser goto /settings/app
@@ -128,8 +129,8 @@ control-openhands browser click testid=submit-button
control-openhands browser viewport phone
control-openhands browser goto "/automations/$QA_HEADER_ID"
control-openhands browser reload
-control-openhands browser screenshot --feature F23.phone --name detail-header-clipping
-control-openhands assert --field value.inside --equals true --bug 'value.inside eq false && value.dir eq "rtl" && value.viewport eq 390 && value.width eq 40' -- browser eval "(()=>{const r=document.querySelector('[role=switch]').getBoundingClientRect();return {inside:r.left>=0&&r.right<=innerWidth,dir:document.documentElement.dir,viewport:innerWidth,width:r.width,left:r.left,right:r.right,scrollWidth:document.documentElement.scrollWidth}})()" # bug
+control-openhands browser screenshot --feature F23.phone --name detail-header-rtl
+control-openhands assert --field value.inside --equals true -- browser eval "(()=>{const r=document.querySelector('[role=switch]').getBoundingClientRect();return {inside:r.left>=0&&r.right<=innerWidth,dir:document.documentElement.dir,viewport:innerWidth,width:r.width,left:r.left,right:r.right,scrollWidth:document.documentElement.scrollWidth}})()"
```
## Gotchas
@@ -147,7 +148,7 @@ control-openhands assert --field value.inside --equals true --bug 'value.inside
- `run-phase` never appeared in the log here (`phase_code` stays `null` on the local service), so the phase chip is not reachable on this stack.
- An LLM profile that does not exist (`model: "qa-no-such-profile"`) does not fail the run: it ran and succeeded with the active profile. Use timeout `1` to get a failed run; it takes several minutes to fail.
- `Load more` in the log (`F23.activity-log-more`) needs more than 20 runs of one automation; at about 35 s and one model call each it is left out of a normal pass (not-run).
-- The script section needs a tarball automation. The account-free way to get one is F22's **Daily news digest** bundle entry (it uploads its bundle and creates the automation); the custom-automation **Upload tarball** action takes `control-openhands fixture tarball`. Its kebab Edit shows generated starter Python instead of the saved bundle (#18341, see F21.edit); a cold load of `/automations/setup?automationId=<id>` shows the **Existing code bundle** summary.
+- The script section needs a tarball automation. The account-free way to get one is F22's **Daily news digest** bundle entry (it uploads its bundle and creates the automation); the custom-automation **Upload tarball** action takes `control-openhands fixture tarball`. Its kebab Edit and a cold load of `/automations/setup?automationId=<id>` both show the **Existing code bundle** summary (see F21.edit, `F23.edit-save`).
- **Starting…** on Run now lasts only for the dispatch request (under 50 ms locally); `--observe` does not catch it.
- The run row is a plain `<a href>`, so opening a run's conversation reloads the whole app (a `document` request in `browser network`), unlike sidebar navigation.
- A 404 detail request is retried three times before **Automation not found** shows (about 7 s of skeleton) ([#17948](https://github.com/OpenHands/OpenHands/issues/17948)); wait with `wait-text`, not `count`.
diff --git a/.agents/skills/verify-openhands/references/feature-map/F25-backends-and-cloud.md b/.agents/skills/verify-openhands/references/feature-map/F25-backends-and-cloud.md
index 1b5227391e..66b285174c 100644
--- a/.agents/skills/verify-openhands/references/feature-map/F25-backends-and-cloud.md
+++ b/.agents/skills/verify-openhands/references/feature-map/F25-backends-and-cloud.md
@@ -22,7 +22,7 @@ Source: `src/components/features/backends/`, `src/contexts/active-backend-contex
## Sub-features
- `F25.selector-dropdown`: clicking (or hovering) the selector opens a listbox of backends, each option named `<status> <name>` with a status dot, and a footer with **Add Backend** and **Manage Backends**.
-- `F25.add-backend-modal`: **Add Backend** opens "Choose how you want to connect" with a description, a **Learn more** link to the deployment docs and the tabs **OpenHands Cloud** (selected) and **Agent-server** (plus **Providers** with the bridge, see the Providers row). It takes keyboard focus on open (the close button), Tab and Shift+Tab wrap inside it, and at desktop width closing returns focus to the backend selector (at phone width the drawer closes with the modal and focus falls back to the page body). Escape does not close it; the X does. Known failure (reproduced 2026-10-17, [#18589](https://github.com/OpenHands/OpenHands/issues/18589)): at phone width, with the modal opened from the drawer's backend selector, Escape closes the drawer and the modal with it.
+- `F25.add-backend-modal`: **Add Backend** opens "Choose how you want to connect" with a description, a **Learn more** link to the deployment docs and the tabs **OpenHands Cloud** (selected) and **Agent-server** (plus **Providers** with the bridge, see the Providers row). It takes keyboard focus on open (the close button), Tab and Shift+Tab wrap inside it, and closing returns focus to the backend selector it was opened from. Escape does not close it; the X does. At phone width, opened from the drawer's backend selector, Escape leaves the modal and the drawer open, and the X closes only the modal (the drawer stays open).
- `F25.cloud-advanced-host`: on the Cloud tab, **Advanced** reveals a **Host** field (placeholder `https://app.all-hands.dev`) for a self-hosted Cloud.
- `F25.cloud-device-flow`: **Connect to OpenHands** opens a popup and shows `Starting authentication...`, then `Waiting for authorization...` with the verification link and **Cancel**; Cancel returns to the idle button and closes the popup. An unreachable host shows `Failed to start device flow: ...` with **Try again** and closes the popup (Try again opens a fresh one); the error text, the `Waiting for authorization...` status, the `Browser opened...` line and the verification link stay readable (at least 4.5:1) in every theme, Light+ and Solarized Light included. Known failure (reproduced 2026-10-17, [#18546](https://github.com/OpenHands/OpenHands/issues/18546)): in Solarized Light the `Browser opened. Complete sign-in to continue...` line (`text-tertiary` `#586E75` on the modal's `#EEE8D5`) measures 4.39:1. Approving the code (adds a backend named OpenHands Cloud) needs a Cloud account.
- `F25.agent-server-guidance`: the Agent-server tab has a **Local / Remote** toggle. Local shows a collapsible **Before you connect** with `agent-canvas --backend-only --port 8001` and a local-setup docs link; Remote shows **Recommended setup**, **Connection details** and a remote-setup docs link.
@@ -30,17 +30,17 @@ Source: `src/components/features/backends/`, `src/contexts/active-backend-contex
- `F25.switch-backend`: choosing another option shows a full-screen `Switching to <name>` overlay for about a second, then the sidebar and pages show that backend's data; on `/conversations/<id>` the app moves to `/conversations` (BM-002).
- `F25.backend-pinned-url`: in-app links carry `?backend=<id>`; opening such a URL pins the tab to that backend; an unknown id is ignored.
- `F25.per-tab-backend`: the active backend is per browser tab: switching it in one tab leaves the other open tabs where they were, a new tab opened from a plain link starts from the last choice made in any tab, and a sidebar conversation row opened in a new tab (Control-click) lands on the backend that owns the conversation because its link carries `?backend=<owner>`.
-- `F25.manage-backends`: Manage backends lists every backend with status dot, name, version (`v1.50.1`), host, status text and a LOCAL/CLOUD pill, plus Edit and Remove; it takes keyboard focus on open and keeps Tab inside; Escape, a backdrop click, X and **Done** close it and return focus to the selector.
-- `F25.manage-add`: **Add Backend** in Manage backends opens the same Add form stacked over Manage; its X returns to Manage, Escape leaves both open (Known failure (reproduced 2026-10-17, [#18589](https://github.com/OpenHands/OpenHands/issues/18589)): at phone width, with Manage opened from the drawer's backend selector, Escape closes the drawer, Add and Manage), and a successful add keeps Manage open with the new row while the new backend becomes active.
+- `F25.manage-backends`: Manage backends lists every backend with status dot, name, version (`v1.50.1`), host, status text and a LOCAL/CLOUD pill, plus Edit and Remove; it takes keyboard focus on open and keeps Tab inside; Escape, a backdrop click, X and **Done** close it and return focus to the selector. At phone width, opened from the drawer, Escape closes only Manage and the drawer stays open.
+- `F25.manage-add`: **Add Backend** in Manage backends opens the same Add form stacked over Manage; its X returns to Manage, Escape leaves both open (at phone width, with Manage opened from the drawer's backend selector, the drawer stays open too), and a successful add keeps Manage open with the new row while the new backend becomes active.
- `F25.manage-select`: clicking a healthy row makes that backend active and closes the modal; on a detail page it leaves it for that section's list, like the selector does (BM-002).
- `F25.edit-backend`: the pencil opens "Edit backend" pre-filled (key masked) with a `Connected · Local v<version>` badge; empty name and invalid host show `Name is required` and `Enter a valid URL (e.g. http://localhost:8080)` and disable Save; Save re-tests the connection, keeps the modal open with `Could not connect to <host>` on failure, and persists on success.
- `F25.edit-cancel`: **Cancel**, the X and a backdrop click in Edit backend close only the Edit modal and discard the draft; Manage backends stays open with the stored values.
-- `F25.edit-escape`: Escape in the Edit modal does nothing: Edit keeps its draft and Manage backends stays open underneath. Known failure (reproduced 2026-10-17, [#18589](https://github.com/OpenHands/OpenHands/issues/18589)): at phone width, with Manage opened from the drawer's backend selector, Escape closes the drawer, Manage and Edit, and the draft is lost.
+- `F25.edit-escape`: Escape in the Edit modal does nothing: Edit keeps its draft and Manage backends stays open underneath; at phone width, with Manage opened from the drawer's backend selector, the drawer stays open too.
- `F25.health-status`: a backend that stops answering turns its dot red (`Disconnected`) within one health probe (every 30 s for local and remote agent servers; Cloud backends every 5 min and not on window focus or reconnect, #18128); its Manage row reads `Disconnected (check URL or network)` with a red detail line and cannot be selected.
- `F25.recovery-gate`: reloading while the active backend is down shows only `agent-server-onboarding-screen` with Manage backends in recovery mode (no X, no Done, Escape ignored, primary **Add Backend**); choosing a healthy row restores the app.
- `F25.remove-backend`: the trash icon asks `Remove backend "<name>"? If it is active, the app will switch back to Local.`; Cancel keeps it, Confirm removes it and an active removed backend falls back to Local (BM-003).
- `F25.collapsed-entry`: with the rail collapsed, the backend icon's popover offers the same Add Backend and Manage Backends items, and both modals open.
-- `F25.phone`: at 390 px the drawer's selector opens Add and Manage inside the viewport without horizontal overflow. Edit backend fits at 320 and 390 px; on short viewports its form scrolls while the title and close control stay visible, and Cancel/Save remain reachable.
+- `F25.phone`: at 390 px the drawer's selector opens Add and Manage inside the viewport without horizontal overflow, and Escape in a backend modal (Add, Manage, Edit, the Remove confirmation) never closes the drawer underneath. Edit backend fits at 320 and 390 px; on short viewports its form scrolls while the title and close control stay visible, and Cancel/Save remain reachable.
- `F25.device-verify-page`: `/oauth/device/verify` shows a "Device Authorization" code form; with `?user_code=` it shows "Device Authorization Request", the code, a Security Notice and Cancel / Authorize Device. On a local stack Authorize and Continue end in the `Error` card with **Try Again**.
- `F25.shared-conversation-view`: `/shared/conversations/<id>` is a standalone read-only page (no app shell); on a local backend it reads `Conversation not found`.
- `F25.share-publicly`: on a Cloud backend the conversation menu has a **Public Share** toggle with copy-link and open-link buttons; on a local backend the item is absent.
@@ -85,21 +85,21 @@ Preconditions:
- **Device flow awaiting and Cancel (`F25.cloud-device-flow`).** Reopen Add Backend (Cloud tab, default host), run `control-openhands browser click 'testid=add-backend-login-button' --observe 'testid=add-backend-device-flow' --observe-ms 6000`, then `control-openhands browser snapshot 'testid=add-backend-device-flow'`. If `https://app.all-hands.dev` is reachable (`control-openhands browser network --external --last 5` lists it), the snapshot shows `Waiting for authorization...`, `Browser opened. Complete sign-in to continue. If browser didn't open, visit:`, a link `https://app.all-hands.dev/oauth/device/verify?user_code=<code>` and button `Cancel`; `browser tabs` lists a second page on that URL. The status line, the `Browser opened...` line and the link stay at least 4.5:1 in Light+ and Solarized Light (link 14.85:1 and 12.25:1, `Browser opened...` 7.62:1 in Light+ on 2026-10-17). Known failure (reproduced 2026-10-17, [#18546](https://github.com/OpenHands/OpenHands/issues/18546)): the `Browser opened...` line measures 4.39:1 in Solarized Light. Screenshot with `--feature F25.cloud-device-flow --name awaiting`, then `control-openhands browser click 'testid=add-backend-auth-cancel'`: the panel is back to `Connect to OpenHands` / `Advanced` and `browser tabs` lists one page. Do not approve the code here (the **Cloud backend** bullet does, when it can). If Cloud is unreachable you get the error state of the previous bullet instead.
- **Agent-server guidance (`F25.agent-server-guidance`).** Run `control-openhands browser click 'testid=add-backend-option-agent-server'` and `control-openhands browser snapshot 'testid=add-backend-agent-server-panel'`: radiogroup `Agent-server location` with `Local` checked, heading `Before you connect`, textboxes `Host Name`, `Host`, `API Key` and `button "Connect" [disabled]`. `control-openhands browser click 'testid=add-backend-local-guidance-toggle'` and `browser snapshot 'testid=add-backend-local-guidance'` show the code `agent-canvas --backend-only --port 8001` and link `Read the local backend setup guide`. `control-openhands browser click 'testid=add-backend-location-option-remote'` and `control-openhands browser click 'testid=add-backend-remote-guidance-toggle'`; `browser snapshot 'testid=add-backend-remote-guidance'` shows `Recommended setup`, `Connection details` and `Read the remote backend setup guide`.
- **Connect rules and errors (`F25.add-agent-server`).** Still on Remote: `control-openhands browser fill 'testid=add-backend-name' QA_Second`, `control-openhands browser fill 'testid=add-backend-host' 127.0.0.1:9`; `control-openhands browser enabled 'testid=add-backend-submit'` is `false` (Remote needs a key). `control-openhands browser fill 'testid=add-backend-api-key' wrong-key` makes it `true`. `control-openhands browser click 'testid=add-backend-location-option-local'` keeps the fields. `control-openhands browser click 'testid=add-backend-submit' --observe 'testid=add-backend-submit'` shows `Checking…`, then `control-openhands browser text 'testid=add-backend-error'` starts `Could not connect to http://127.0.0.1:9` / `Disconnected (check URL or network). ...`. Fill `add-backend-host` with `<second-url>` and submit again: the error reads `Could not connect to <second-url>` / `Invalid API key`. On Local the key is optional for Connect but not for a key-protected host: `control-openhands browser fill 'testid=add-backend-api-key' ''` leaves `control-openhands browser enabled 'testid=add-backend-submit'` at `true` (`browser value 'testid=add-backend-api-key'` reports `length` `0`), and `control-openhands browser click 'testid=add-backend-submit'` then `control-openhands browser wait 'testid=add-backend-error' --timeout 15000` shows the same `Could not connect to <second-url>` / `Invalid API key` (`browser screenshot --feature F25.add-agent-server --name local-no-key-error`). Close with `testid=add-backend-close`.
-- **Add the second backend (`F25.add-agent-server`).** Run `control-openhands browser goto /conversations/<id>`, `browser click 'testid=backend-selector'`, `browser click 'testid=add-backend-menu-item'`, `browser click 'testid=add-backend-option-agent-server'`, fill `add-backend-name` with `QA_Second` and `add-backend-host` with `<second-url>`, then `control-openhands browser fill 'testid=add-backend-api-key' --value-file "$OH_VERIFY_RUN_2/private/session-key"` and `control-openhands browser click 'testid=add-backend-submit' --expect-url '/conversations(\?|$)'`. The URL is `/conversations`, `browser count 'testid=add-backend-modal'` is `0`, `browser snapshot 'testid=backend-selector'` shows `combobox "QA_Second"`, and `browser count 'testid=conversation-card'` is `0`. Before anything reloads the page, `control-openhands browser toasts --history` lists only `Loading...` and the new backend's `Your LLM isn't set up yet, so conversations won't run. ...` (the second stack has no LLM profile; no error toast), and `control-openhands browser network --external --filter 'ids='` lists no request: the new backend is never asked for the first stack's conversation id. Known failure (reproduced 2026-10-13 and 2026-10-17, #18390): on the run's first activation of `QA_Second` the list sometimes holds `GET <second-url>/api/conversations?ids=<id>` (`200`); record the bullet as `fail` then (see Gotchas). (`--external` leaves out this stack's own `GET /api/conversations?ids=<id>` from the `goto /conversations/<id>` above, which the request log keeps across page loads; `<second-url>` is another origin, so a request to it is listed.) The new backend asks its own telemetry consent ("This preference is saved for the local backend “QA_Second”"): run `control-openhands onboard --skip` (it leaves the browser on `/`). After `control-openhands browser reload` the selector still reads `QA_Second`.
+- **Add the second backend (`F25.add-agent-server`).** Run `control-openhands browser goto /conversations/<id>`, `browser click 'testid=backend-selector'`, `browser click 'testid=add-backend-menu-item'`, `browser click 'testid=add-backend-option-agent-server'`, fill `add-backend-name` with `QA_Second` and `add-backend-host` with `<second-url>`, then `control-openhands browser fill 'testid=add-backend-api-key' --value-file "$OH_VERIFY_RUN_2/private/session-key"` and `control-openhands browser click 'testid=add-backend-submit' --expect-url '/conversations(\?|$)'`. The URL is `/conversations`, `browser count 'testid=add-backend-modal'` is `0`, `browser snapshot 'testid=backend-selector'` shows `combobox "QA_Second"`, and `browser count 'testid=conversation-card'` is `0`. Before anything reloads the page, `control-openhands browser toasts --history` lists only `Loading...` and the new backend's `Your LLM isn't set up yet, so conversations won't run. ...` (the second stack has no LLM profile; no error toast), and `control-openhands browser network --external --filter 'ids='` lists no request: the new backend is never asked for the first stack's conversation id. Known failure (reproduced 2026-10-13, 2026-10-17 and 2026-10-19, #18390): on the run's first activation of `QA_Second` the list sometimes holds `GET <second-url>/api/conversations?ids=<id>` (`200`); record the bullet as `fail` then (see Gotchas). (`--external` leaves out this stack's own `GET /api/conversations?ids=<id>` from the `goto /conversations/<id>` above, which the request log keeps across page loads; `<second-url>` is another origin, so a request to it is listed.) The new backend asks its own telemetry consent ("This preference is saved for the local backend “QA_Second”"): run `control-openhands onboard --skip` (it leaves the browser on `/`). After `control-openhands browser reload` the selector still reads `QA_Second`.
- **Switch with the overlay (`F25.switch-backend`).** From `/` run `browser click 'testid=backend-selector'` and `control-openhands browser click 'testid=backend-selector >> role=option[name="Connected Local"]' --observe 'testid=environment-switch-overlay' --observe-ms 2500`. The observation shows `Switching to Local`, then `<absent>` about one second later; the selector reads `Local` and `browser count 'testid=conversation-card'` is `1`. Then `control-openhands browser click 'testid=conversation-card' --expect-url '/conversations/[0-9a-f-]+'`, open the selector and `control-openhands browser click 'testid=backend-selector >> role=option[name="Connected QA_Second"]' --expect-url '/conversations(\?|$)'`: the URL is `/conversations`, the selector reads `QA_Second` and the card count is `0`.
- **Pinned URL (`F25.backend-pinned-url`).** With `QA_Second` active run `control-openhands browser goto '/conversations/<id>?backend=default-local'` and `control-openhands browser wait 'testid=chat-interface'`: the selector reads `Local` and the conversation opens. `control-openhands browser goto '/conversations?backend=qa-unknown-id'` keeps `Local` (card count `1`). The sidebar card links themselves end in `?backend=default-local` (`browser snapshot`).
- **Per-tab backend (`F25.per-tab-backend`).** With `Local` active and `QA_Second` registered, run `control-openhands browser goto /conversations` and `control-openhands browser eval "JSON.parse(sessionStorage.getItem('openhands-active-backend')).backendId"` (`default-local`); note QA_Second's id with `control-openhands browser eval "JSON.parse(localStorage.getItem('openhands-backends')).find(b=>b.name==='QA_Second').id"` (`<second-id>`, a generated UUID). Open a plain second tab, as a user opening the app again: `control-openhands browser tab new /conversations` prints `index` `1` and makes it the daemon's active tab (`control-openhands browser tabs` lists two pages at `/conversations` with `active` `1`); `control-openhands browser wait-tab '/conversations$' --new --timeout 5000` then prints `index` `0`, the tab you came from, because `--new` leaves out the tab the daemon is on and that is now the new one (run from `browser tab 0` it prints `index` `1`, the second tab). `control-openhands browser tab 1`, `control-openhands browser wait 'testid=backend-selector'`: `browser snapshot 'testid=backend-selector'` reads `combobox "Local"` and `control-openhands browser eval "sessionStorage.getItem('openhands-active-backend')"` is `null` (a new tab inherits nothing and starts from the localStorage fallback). Switch this tab: `browser click 'testid=backend-selector'`, `control-openhands browser click 'testid=backend-selector >> role=option[name="Connected QA_Second"]'`, `control-openhands browser wait 'testid=environment-switch-overlay' --state detached --timeout 10000`; the selector reads `QA_Second`, `browser count 'testid=conversation-card'` is `0` and `control-openhands browser eval "JSON.parse(localStorage.getItem('openhands-active-backend')).backendId"` is `<second-id>`. Back in `control-openhands browser tab 0` nothing moved: the selector still reads `Local`, the card count is `1`, the sessionStorage eval is still `default-local` while the localStorage eval is `<second-id>` (`browser screenshot --feature F25.per-tab-backend --name first-tab-unchanged`). A plain new tab starts from the last choice made in any tab: from tab 0 run `control-openhands browser tab new /conversations` (`index` `2`, now active), `browser wait 'testid=backend-selector'`: the selector reads `QA_Second`, the card count is `0`, its sessionStorage item is `null` and the localStorage eval is `<second-id>` (`--name plain-new-tab-last-choice`). A conversation row opens on its owner instead: `control-openhands browser tab 0`, `control-openhands browser click 'testid=conversation-card' --modifiers Control` (add `>> nth=0` when the run holds more conversations than the precondition's one: the click is strict), `control-openhands browser wait-tab '/conversations/<id>' --timeout 20000` prints `index` `3` and the URL `/conversations/<id>?backend=default-local`; `control-openhands browser tab 3`, `control-openhands browser wait 'testid=chat-interface' --timeout 30000`, `browser url` still ends in `/conversations/<id>?backend=default-local` (no bounce to `/conversations`), the selector reads `Local` and `browser toasts --history` lists only `Loading...` (no `This conversation does not exist…`); `browser screenshot --feature F25.per-tab-backend --name new-tab-on-owner`. Close the extra tabs highest first: `control-openhands browser close-tab 3`, `control-openhands browser close-tab 2`, `control-openhands browser close-tab 1`, `control-openhands browser tab 0`; `browser tabs` lists one page and the selector reads `Local`. Opening the pinned link also recorded `default-local` as the last choice (the localStorage eval in tab 3 reads `default-local`), so run the plain-new-tab check before the row check.
- **Manage list and closing (`F25.manage-backends`).** Run `browser click 'testid=backend-selector'`, `control-openhands browser click 'testid=manage-backends-menu-item'` and `control-openhands browser snapshot 'testid=manage-backends-modal'`: heading `Manage backends`, two rows `Connected Local v1.50.1 <this run's URL> Connected Local` and `Connected QA_Second v1.50.1 <second-url> Connected Local`, each with `Edit` and `Remove`, then `Add Backend` and `Done`. `browser text 'testid=manage-backends-status-QA_Second'` is `Connected` and `browser text 'testid=manage-backends-version-QA_Second'` is `v1.50.1` (whatever `doctor` reports). Screenshot with `--feature F25.manage-backends --name list`. Focus is inside: the Add Backend bullet's focus eval with `manage-backends-modal` in place of `add-backend-modal` reads `close-manage-backends-modal` on open and stays `inDialog` `true` over nine Tabs and two Shift+Tabs (with two backends listed, Tab 8 reaches **Done** and Tab 9 wraps to the close button); after Escape, focus is back on the selector. Close it four ways, reopening between: `browser press Escape`, `control-openhands browser mouse-click 20 500` (backdrop), `browser click 'testid=close-manage-backends-modal'`, `browser click 'testid=manage-backends-done'`; each time `browser count 'testid=manage-backends-modal'` is `0`.
- **Select from Manage (`F25.manage-select`).** With `Local` active, `control-openhands browser click 'testid=conversation-card' --expect-url '/conversations/[0-9a-f-]+'`, open Manage Backends, then `control-openhands browser click 'testid=manage-backends-row-QA_Second >> role=button >> nth=0' --expect-url '/conversations(\?|$)'`. Like the selector, the app leaves the detail page for `/conversations` (BM-002): `browser count 'testid=manage-backends-modal'` is `0`, the selector reads `QA_Second`, `browser count 'testid=chat-interface'` and `browser count 'testid=conversation-card'` are `0`, and `browser screenshot --feature F25.manage-select --name after-select` shows QA_Second's Home with its empty conversation list.
- **Edit validation and errors (`F25.edit-backend`).** Open Manage Backends, `control-openhands browser click 'testid=manage-backends-edit-QA_Second'`, then `browser value 'testid=edit-backend-host'` (`<second-url>`), `browser attr 'testid=edit-backend-api-key' type` (`password`) and `browser text 'testid=edit-backend-status'` (`Connected`, `·`, `Local`, `v1.50.1` on separate lines). Fill `edit-backend-name` with an empty string and `edit-backend-host` with `not a host`, then `browser focus 'testid=edit-backend-api-key'` and `browser snapshot 'testid=edit-backend-form'`: alerts `Name is required` and `Enter a valid URL (e.g. http://localhost:8080)`; `browser enabled 'testid=edit-backend-submit'` is `false`.
-- **Escape in Edit (`F25.edit-escape`).** In the same Edit modal run `control-openhands browser fill 'testid=edit-backend-name' QA_Draft` and `control-openhands browser press Escape`: `browser count 'testid=edit-backend-modal'` and `browser count 'testid=manage-backends-modal'` both stay `1` and `browser value 'testid=edit-backend-name'` is still `QA_Draft` (`browser screenshot --feature F25.edit-escape --name after-escape`). Leave with `browser click 'testid=edit-backend-cancel'`; nothing was saved (the Manage row is still `QA_Second`).
+- **Escape in Edit (`F25.edit-escape`).** In the same Edit modal run `control-openhands browser fill 'testid=edit-backend-name' QA_Draft` and `control-openhands browser press Escape`: `browser count 'testid=edit-backend-modal'` and `browser count 'testid=manage-backends-modal'` both stay `1` and `browser value 'testid=edit-backend-name'` is still `QA_Draft` (`browser screenshot --feature F25.edit-escape --name after-escape`). Leave with `browser click 'testid=edit-backend-cancel'`; nothing was saved (the Manage row is still `QA_Second`). At phone width (the **Phone** bullet's drawer → Manage Backends → `manage-backends-edit-Local`), the same fill and Escape leave `edit-backend-modal`, `manage-backends-modal` and `sidebar-mobile-drawer` at count `1` with the draft kept (`--name phone-after-escape`).
- **Edit and save (`F25.edit-backend`).** Manage Backends is still open: click the pencil `testid=manage-backends-edit-QA_Second` again. Fill `edit-backend-host` with `http://127.0.0.1:9`, `browser click 'testid=edit-backend-submit'`, `browser wait 'testid=edit-backend-error' --timeout 10000`: the text starts `Could not connect to http://127.0.0.1:9` and the modal stays. Fill the host back with `<second-url>` and `edit-backend-name` with `QA_Renamed`, click `edit-backend-submit`, then `control-openhands browser wait 'testid=edit-backend-modal' --state detached --timeout 10000`; `browser count 'testid=manage-backends-row-QA_Renamed'` is `1`. Click `manage-backends-done`, `browser reload`: the selector reads `QA_Renamed`.
- **Health status (`F25.health-status`).** With QA_Renamed the active backend (as the Edit bullet's reload leaves it; the closed selector shows only the active backend's dot), run `control-openhands service stop agent-server --run "$OH_VERIFY_RUN_2"`, then on the open page `control-openhands browser wait 'testid=backend-selector >> role=status[name="Disconnected"]' --timeout 40000`. `browser click 'testid=backend-selector'` and `browser snapshot 'testid=backend-selector'` show `option "Disconnected QA_Renamed"` next to `option "Connected Local"`. In Manage Backends `browser text 'testid=manage-backends-status-QA_Renamed'` is `Disconnected (check URL or network)`, `testid=manage-backends-status-detail-QA_Renamed` repeats it with `Check that the backend URL is correct ...`, and `browser enabled 'testid=manage-backends-row-QA_Renamed >> role=button >> nth=0'` is `false`. An error toast with the same text also appears. Screenshot with `--feature F25.health-status --name manage-down`.
- **Recovery gate (`F25.recovery-gate`).** Close the modal, `control-openhands browser reload`, then `control-openhands browser wait 'testid=manage-backends-modal' --timeout 20000`. `browser testids` lists `agent-server-onboarding-screen` and the Manage rows but no sidebar; `browser count 'testid=close-manage-backends-modal'` and `browser count 'testid=manage-backends-done'` are `0`, and `browser press Escape` leaves the modal (`count` `1`). `control-openhands browser click 'testid=manage-backends-row-Local >> role=button >> nth=0'` and `browser wait 'testid=agent-server-onboarding-screen' --state detached --timeout 15000` restore the app: selector `Local`, card count `1`. Screenshot the gate with `--feature F25.recovery-gate --name gate` before choosing.
- **Remove the active backend (`F25.remove-backend`).** Run `control-openhands restart --run "$OH_VERIFY_RUN_2"`, `browser click 'testid=backend-selector'`, `browser wait 'testid=backend-selector >> role=option[name="Connected QA_Renamed"]' --timeout 25000` and click that option (selector reads `QA_Renamed`). Open Manage Backends, `control-openhands browser click 'testid=manage-backends-remove-QA_Renamed'`; `browser text 'testid=confirmation-modal'` reads `Remove backend "QA_Renamed"? If it is active, the app will switch back to Local.` `control-openhands browser click 'testid=confirmation-modal >> role=button[name="Cancel"]'` keeps the row (`count 'testid=manage-backends-row-QA_Renamed'` `1`). Click the trash again, then `control-openhands browser click 'testid=confirmation-modal >> testid=confirm-button'`: the row count is `0`. Click Done and `browser reload`: the selector reads `Local`, `browser count 'testid=backend-selector >> role=option'` (dropdown open) is `1`, and the card count is `1`.
- **Add from Manage, Edit Cancel and X (`F25.manage-add`, `F25.edit-cancel`).** The second stack is running again after the previous bullet. Run `browser goto /conversations/<id>`, `browser wait 'testid=chat-interface'`, open the selector, click `manage-backends-menu-item`, then `control-openhands browser click 'testid=manage-backends-add'` and `control-openhands browser wait 'testid=add-backend-modal' --timeout 5000` (`visible`); `browser count 'testid=manage-backends-modal'` is `1` underneath. `control-openhands browser press Escape` leaves both (`add-backend-modal` and `manage-backends-modal` counts `1`); `browser click 'testid=add-backend-close'`: the Add count is `0` and the Manage count stays `1`. Click `manage-backends-add` again, `add-backend-option-agent-server`, fill `add-backend-name` with `QA_Second`, `add-backend-host` with `<second-url>` and `add-backend-api-key` with `--value-file "$OH_VERIFY_RUN_2/private/session-key"`, then `browser click 'testid=add-backend-submit' --expect-url '/conversations(\?|$)'`: the URL is `/conversations`, `count 'testid=add-backend-modal'` is `0`, `count 'testid=manage-backends-modal'` is still `1`, `count 'testid=manage-backends-row-QA_Second'` is `1` and the selector reads `QA_Second` (screenshot `--feature F25.manage-add --name after-add`). No consent prompt this time: it is stored on the second backend. Then `control-openhands browser click 'testid=manage-backends-edit-QA_Second'`, `browser fill 'testid=edit-backend-name' QA_Draft` and `control-openhands browser click 'testid=edit-backend-cancel'`: `count 'testid=edit-backend-modal'` `0`, `count 'testid=manage-backends-modal'` `1`, row `QA_Second` `1`, row `QA_Draft` `0`. Repeat with `control-openhands browser click 'testid=edit-backend-close'` and with a backdrop click (`control-openhands browser mouse-click 20 500`): same counts, and reopening the pencil shows `browser value 'testid=edit-backend-name'` `QA_Second`; leave with `edit-backend-cancel`. Remove it again: `browser click 'testid=manage-backends-remove-QA_Second'`, `browser click 'testid=confirmation-modal >> testid=confirm-button'`, `browser click 'testid=manage-backends-done'` and `browser reload`: the selector reads `Local`.
- **Collapsed rail (`F25.collapsed-entry`).** Run `control-openhands browser click 'testid=sidebar-collapse-toggle'` (`attr 'aside[data-collapsed]' data-collapsed` is `true`), `control-openhands browser hover 'testid=collapsed-backend-selector-link'`, `control-openhands browser click 'testid=add-backend-menu-item'` and `control-openhands browser wait 'testid=add-backend-modal' --timeout 5000` (`visible`). Close it, hover again, click `manage-backends-menu-item` and `browser wait 'testid=manage-backends-modal' --timeout 5000`. Close with Done and expand with `browser click 'testid=sidebar-collapse-toggle'`.
-- **Phone (`F25.phone`).** Run `control-openhands browser viewport phone`, `browser goto /conversations`, `control-openhands browser click 'testid=sidebar-mobile-menu-toggle'`, `control-openhands browser click 'testid=sidebar-mobile-drawer >> testid=backend-selector'`, `browser click 'testid=add-backend-menu-item'`, `browser click 'testid=add-backend-option-agent-server'` and `control-openhands browser bbox 'testid=add-backend-modal'`: `insideViewport` `true`, `pageHorizontalOverflow` `false`. Screenshot `--feature F25.phone --name add-modal`. Close it (the drawer stays open), open Manage Backends from the drawer selector the same way and check `bbox 'testid=manage-backends-modal'` the same. Click `testid=manage-backends-edit-Local` and check `bbox 'testid=edit-backend-modal'`: `insideViewport` `true`; repeat at `browser viewport 320x568`. At `browser viewport 320x400`, clear `edit-backend-name`, fill `edit-backend-host` with `not a host` and focus `edit-backend-api-key` to show both validation messages. Run `browser scroll 'testid=edit-backend-form' --by 600`: Cancel and Save must come into view while the title and `edit-backend-close` stay visible. Check the modal and all three controls with `bbox`, then close with Cancel. Return with `browser viewport desktop`; reopen Edit and confirm its width is still 520px.
+- **Phone (`F25.phone`).** Run `control-openhands browser viewport phone`, `browser goto /conversations`, `control-openhands browser click 'testid=sidebar-mobile-menu-toggle'`, `control-openhands browser click 'testid=sidebar-mobile-drawer >> testid=backend-selector'`, `browser click 'testid=add-backend-menu-item'`, `browser click 'testid=add-backend-option-agent-server'` and `control-openhands browser bbox 'testid=add-backend-modal'`: `insideViewport` `true`, `pageHorizontalOverflow` `false`. Screenshot `--feature F25.phone --name add-modal`. Before closing it, `control-openhands browser press Escape` leaves `browser count 'testid=add-backend-modal'` and `browser count 'testid=sidebar-mobile-drawer'` at `1`. Close it with `testid=add-backend-close`: the drawer stays open and `control-openhands browser eval "!!document.activeElement.closest('[data-testid=sidebar-mobile-drawer] [data-testid=backend-selector]')"` is `true`. Open Manage Backends from the drawer selector the same way and check `bbox 'testid=manage-backends-modal'` the same. Escape inside a stacked modal reaches only that modal: `manage-backends-add` then Escape keeps Add, Manage and the drawer (`--feature F25.manage-add --name phone-escape-stays`), and its X returns to Manage; `manage-backends-remove-Local` opens the Remove confirmation, and Escape closes only it (Manage, the drawer and the `Local` row stay; `--name remove-confirm-escape-local`). Escape on Manage itself closes only Manage (drawer count `1`); a further Escape then lands on the focused drawer selector (see [#18233](https://github.com/OpenHands/OpenHands/issues/18233) in Gotchas), so reopen Manage from the selector before the Edit step. Click `testid=manage-backends-edit-Local` and check `bbox 'testid=edit-backend-modal'`: `insideViewport` `true`; repeat at `browser viewport 320x568`. At `browser viewport 320x400`, clear `edit-backend-name`, fill `edit-backend-host` with `not a host` and focus `edit-backend-api-key` to show both validation messages. Run `browser scroll 'testid=edit-backend-form' --by 600`: Cancel and Save must come into view while the title and `edit-backend-close` stay visible. Check the modal and all three controls with `bbox`, then close with Cancel. Return with `browser viewport desktop`; reopen Edit and confirm its width is still 520px.
- **Providers tab (`F25.providers-tab`).** On the default run `browser eval "!!window.marsBridge"` is `true` (Preconditions); `curl -s "http://127.0.0.1:$(control-openhands status | jq -r .ports.frontend)/mars/health"` (the static child behind the ingress) returns the HTML index, not JSON: one bridge per stack. Open Add Backend: the tablist adds `tab "Providers Connect through DigitalOcean and more."` (`testid=add-backend-option-providers`); `control-openhands browser click 'testid=add-backend-option-providers'` and `browser snapshot 'testid=add-backend-providers-panel'` show `button "DigitalOcean Your team's OpenHands agents"` and `More providers coming soon.` (`browser screenshot --feature F25.add-backend-modal --name with-providers-tab`). At phone width the modal stays inside the viewport with the three tab titles cut to `Ope…`, `Age…`, `Pro…` (`--feature F25.providers-tab --name phone-tabs`).
- **DigitalOcean sign-in (`F25.digitalocean-sign-in`).** On the default run, run `control-openhands browser click 'testid=add-backend-provider-digitalocean'` and `browser wait 'testid=managed-agents-sign-in'`: heading `Connect your DigitalOcean team`, `textbox "Personal access token"` (`browser attr 'testid=managed-agents-token' type` is `password`), link `Create a token` (`/url: https://cloud.digitalocean.com/account/api/tokens`) and `button "Connect with token" [disabled]`; no OAuth button in a browser (`browser count 'testid=managed-agents-oauth'` is `0`). Fill `testid=managed-agents-token` with `dop_v1_qa_invalid_token_0000` and click `testid=managed-agents-token-submit`: `browser text 'testid=managed-agents-auth-error'` reads `That does not look like a DigitalOcean personal access token.` (`POST /mars/rpc/savePat` `500`). Fill a well-formed dummy, `dop_v1_` plus 64 zeros, and submit again: after a few seconds the error reads `401 Unauthorized` (`POST /mars/rpc/savePat` `401`, DigitalOcean's answer; `browser screenshot --feature F25.digitalocean-sign-in --name rejected-token`). Never type a real token here. `control-openhands browser click 'testid=add-backend-providers-back'` returns to the Providers panel (`browser count 'testid=managed-agents-sign-in'` is `0`; the providers panel itself stays mounted behind the sign-in screen, so its count is `1` either way); close with `testid=add-backend-close`. At phone width the sign-in screen fits (`bbox 'testid=add-backend-modal'` `insideViewport` `true`). Both errors also raise an error toast with the same text. Expected on a page served without a baked key (F26 `--host 0.0.0.0`, Docker), once the server's key was typed on the API-key screen: the same two answers as above. Known failure (reproduced 2026-10-14, #18434): every token, even the malformed one, ends in `session key required` (`POST /mars/rpc/getAuthState` and `savePat` `401`), because the web bridge client sends only the baked page key, never the typed one; it predates #18349 (BASE with `MARS_WEB=1` fails the same way). A `--frontend-only` page connected to another stack's backend fails the same way for a different reason (the typed key belongs to the other stack); #18434 leaves that case to a maintainer decision.
- **Managed Agents with an account (`F25.digitalocean-agents`, `F25.digitalocean-backend-removal`).** Blocked: needs a DigitalOcean account with Managed Agents. With one, after a real token the panel lists agents (`digitalocean-agent-<id>`), **New Agent** opens `digitalocean-new-agent-form` (name, size), a session's Connect adds a backend and the sidebar shows `mars-session-panel`; removing that backend opens `mars-backend-removal-modal` with `disconnect-only-button` and `stop-session-button`.
@@ -148,7 +148,7 @@ Preconditions:
- Local and remote backends are probed every 30 s (settings + server_info), Cloud backends every 5 min (#18128), so allow up to 40 s for a local dot to change. `browser network --clear`, wait about 70 s, then `browser network --filter server_info` shows two probes per local origin 30 s apart. While the active backend is down, any reload lands on the recovery gate: drive the disconnected rows on the page that was already loaded.
- `/oauth/device/verify` is a Cloud page that also ships in Canvas: on a local stack its POST target does not exist, so Authorize always ends in `Error`. The "Authentication Required" state is unreachable (`useIsAuthed` always resolves `true`).
- The add bullet's `network --external --filter 'ids='` check guards an earlier race: [#18390](https://github.com/OpenHands/OpenHands/issues/18390) tracks the same first-activation symptom; on a backend's first activation from `/conversations/<id>`, about 1 attempt in 5 asked the new backend for the old conversation id. It did not happen in 15 first activations at 4eff80237: each time `QA_Second` was removed, its consent unset with `control-openhands api PATCH /api/settings --run "$OH_VERIFY_RUN_2" --write --data '{"misc_settings_diff":{"app_preferences":{"user_consents_to_analytics":null}}}'` (arrange only) and the backend added again from `/conversations/<id>`. A backend that lacks an id answers `[null]`, which now reads as "not found" (only the `This conversation does not exist…` toast), not as incompatible data. No merged change was shown to fix the race itself (#18009 only changed how a `[null]` answer reads; #18024 only moved the redirect into a shared hook, still called after `addBackend`), so a hit (an `ids=` request to `<second-url>`) is a bug to file. At 95115e84f6dd it hit on 1 of 2 first activations from `/conversations/<id>` (one `GET <second-url>/api/conversations?ids=<id>` answered `200`, with no `This conversation does not exist…` toast after it). The hit followed a failed Connect in the same modal (an empty key, `Invalid API key`) before the successful one; the clean retry, which connected once, had none, so a repro should try both sequences. On 2026-10-08 at d2c89252d it did not reproduce in 17 first activations (8 clean, 9 failed-then-clean, some under CPU load and some human-paced): every `ids=` capture was empty, and no merged change touched the activation logic since the hit, so it stays a watch item rather than a filed defect. Record the Add bullet as `fail` when the request appears, even though the redirect and the selector look right. Known failure (reproduced 2026-10-13, #18390): two independent first activations hit it that day, each the run's first activation of `QA_Second` after a clean Connect. On 2026-10-13 at 2bdc0e9d6 it hit on 1 of 6 first activations from `/conversations/<id>`: the first one of the run (`GET <second-url>/api/conversations?ids=<id>` `200`, no toast), a clean Connect in a modal opened after earlier failed Connects in other Add Backend openings; the next 5 (2 failed-then-clean, 3 clean, plus one Manage → Add) had none.
-- Add Backend and Manage Backends take and keep keyboard focus, and Edit backend takes it on open (`ModalBackdrop`; at phone width open the selector with `'testid=sidebar-mobile-drawer >> testid=backend-selector'`, since the bare test id matches two selectors). At desktop width focus returns to the selector input when they close (at phone width Add Backend's X closes the drawer too and focus falls to the page body), so an Escape pressed after closing lands on the selector and can trigger [#18233](https://github.com/OpenHands/OpenHands/issues/18233) below.
+- Add Backend and Manage Backends take and keep keyboard focus, and Edit backend takes it on open (`ModalBackdrop`; at phone width open the selector with `'testid=sidebar-mobile-drawer >> testid=backend-selector'`, since the bare test id matches two selectors). Focus returns to the selector input when they close (at phone width the drawer's selector, with the drawer still open), so an Escape pressed after closing lands on the selector and can trigger [#18233](https://github.com/OpenHands/OpenHands/issues/18233) below.
- Toasts are not page or HTTP errors, so `browser errors --app-only` misses them: read `browser toasts --history` after every add or switch, before anything reloads the page (the history starts with each page load).
- `F25.locked-cloud` needs `scripts/static-server.mjs --lock-to-cloud <url>`, which neither `bin/agent-canvas.mjs` nor `control-openhands launch` can pass today; even with it, reconnecting needs a Cloud account.
- The Cloud tab title truncates to `OpenHan…` at 390 px when the Agent-server tab is selected; the layout still fits. With the Providers tab present all three titles are cut to three letters.
diff --git a/.agents/skills/verify-openhands/references/feature-map/F26-runtime-variants.md b/.agents/skills/verify-openhands/references/feature-map/F26-runtime-variants.md
index e7452b2ad9..b326b45d34 100644
--- a/.agents/skills/verify-openhands/references/feature-map/F26-runtime-variants.md
+++ b/.agents/skills/verify-openhands/references/feature-map/F26-runtime-variants.md
@@ -40,7 +40,7 @@ Source: `bin/agent-canvas.mjs`, `scripts/dev-with-automation.mjs`, `scripts/dev-
- `F26.runtime-services-agent-use`: with that block alone, the agent can reach the automation backend from its terminal: asked to, it reads the OpenAPI at the URL the block names, creates an enabled cron automation, and dispatches a run that completes; the automation and its run show up in the Automations UI.
- `F26.lib-build`: `npm run build:lib` produces `dist/` where every `package.json` `exports` target exists and the main entry exports `AgentServerUIProviders`, `AgentServerUIRoot`, `CloudOrganizationBoundary` and the telemetry helpers.
- `F26.lib-style-scope`: all bundled CSS is scoped under `[data-agent-server-ui]`, and theme tokens such as `--oh-color-base` live on that scope root (observable in the standalone app).
-- `F26.lib-host-app`: mounted in a separate host app, Canvas styles stay inside the scope and `styleOverrides` restyle it. Known failure (reproduced 2026-10-17, [#18615](https://github.com/OpenHands/OpenHands/issues/18615)): a real Vite 8.0.16 browser host using the normal bare package import and matching React 19.3.0 peers serves HTTP 200 but cannot mount `AgentServerUIRoot`; emitted ESM invokes CommonJS `require("react")` through the rolldown runtime and throws. Export enumeration and CSS scoping pass separately and do not prove host mounting.
+- `F26.lib-host-app`: the built ESM library mounts in a separate Vite browser host that imports `@openhands/agent-canvas` by bare package name with matching React peers: `AgentServerUIRoot` renders its children and they stay interactive, with no module error. Canvas styles apply inside the `[data-agent-server-ui]` scope only and `styleOverrides` restyle that subtree. Known failure (reproduced 2026-10-19, [#18655](https://github.com/OpenHands/OpenHands/issues/18655)): host elements stay unaffected and `styleOverrides` set the variable on the scope root, but the library ships no Canvas utility or theme stylesheet (`dist/` holds only the 5 KB xterm CSS and no JS imports a stylesheet), so a mounted Canvas component (`EmptyBrowserMessage`) renders unstyled and the override has no visible effect.
- `F26.docker-image`: `docker run ghcr.io/openhands/agent-canvas` serves Canvas at `/canvas`; without `AGENT_CANVAS_ALLOW_LAN_SESSION_KEY=true` the UI asks for the key. Blocked without a Docker daemon. On a host whose kernel has no IPv6 (no `/proc/net/if_inet6`) the entrypoint logs `IPv6 is not available; binding the frontend to 0.0.0.0.` and the container keeps running (#18180).
- `F26.docker-conversation-runtime`: `OH_CONVERSATION_RUNTIME=docker` runs each new conversation in its own container. Blocked without a Docker daemon.
- `F26.helm-chart`: `helm install agent-canvas ./helm/agent-canvas` runs the Docker image as a Kubernetes StatefulSet with a PVC and an Ingress. Blocked without `helm` and a cluster.
@@ -96,8 +96,9 @@ Preconditions:
- **Runtime services (`F26.runtime-services`).** Run `control-openhands api GET /server_info --pick runtime_services.mode` (`agent-canvas`) and `control-openhands api GET /server_info --pick runtime_services.services.automation.url_from_agent` (`http://localhost:<automation port>`; `control-openhands status` prints the ports). Then `control-openhands conversation start --prompt "Reply with the single word: ok" --wait --timeout 180` (prints `<id>`) and `control-openhands conversation events <id> --grep RUNTIME_SERVICES --from-start`: the `SystemPromptEvent` matches, with an excerpt `<RUNTIME_SERVICES>\nYou are running inside an agent-canvas dev stack started ...`. `control-openhands conversation events <id> --grep "localhost:<automation port>" --from-start` shows `* Automation backend: http://localhost:<port>` plus its `Docs:` and `OpenAPI:` URLs.
- **The agent uses the runtime services (`F26.runtime-services-agent-use`).** Write `qa-f26-agent.txt` (outside the workspace) containing: ``Using the Automation backend described in your RUNTIME_SERVICES (read its OpenAPI to find the routes), create one automation named QA_rt_auto from a prompt: the prompt is "Reply with the single word: pong. Do not run any tools.", the schedule is the cron 0 9 * * *, and it is enabled. Then dispatch one run of it now. Do not change anything else. Reply with the automation id and the run id only.`` Run `control-openhands conversation start --prompt "$(cat qa-f26-agent.txt)" --wait --timeout 400` (note `<agent-id>`); the reply names `<automation-id>` and a run id. `control-openhands conversation events <agent-id> --kinds ActionEvent --from-start` shows the agent's own `curl` calls: `http://localhost:<automation port>/api/automation/openapi.json`, `POST …/api/automation/v1/preset/prompt` and `POST …/api/automation/v1/<automation-id>/dispatch`, each with the `X-Session-API-Key: $OPENHANDS_AUTOMATION_API_KEY` header the block names. Second view in the UI: `control-openhands browser goto /automations` and `control-openhands browser text '[data-testid^=automation-card-] >> has-text=QA_rt_auto'` (`QA_rt_auto`, the prompt, `cron`, `Runs 1`). Open it with `control-openhands browser click '[data-testid^=automation-card-] >> has-text=QA_rt_auto >> text=QA_rt_auto' --expect-url '/automations/[0-9a-f-]+'`. Within about a minute, `control-openhands browser text 'testid=automation-activity-log'` shows the run's summary (`Replied with the single word "pong" …`), a cost and `Successful` (`control-openhands browser screenshot --feature F26.runtime-services-agent-use --name agent-created-automation-run`). `control-openhands api GET /api/automation/v1/<automation-id>` has `name` `QA_rt_auto`, `trigger.schedule` `0 9 * * *` and `enabled` `true`. Delete it afterwards (`control-openhands api DELETE /api/automation/v1/<automation-id> --write`, arrange). It runs daily at 09:00 UTC otherwise.
-- **Library build (`F26.lib-build`).** Never build inside the shared checkout (it regenerates i18n and typegen files the other runs hash); build in a copy: `L=$OH_VERIFY_RUN/private/f26-lib; mkdir -p $L; tar --exclude=./node_modules --exclude=./build --exclude=./.git --exclude=./dist -cf - . | tar -xf - -C $L; ln -s "$PWD/node_modules" $L/node_modules; (cd $L && npm run build:lib); echo "exit=$?"`. It ends with `✓ built in` and `exit=0` (about 30 s). Then `(cd $L && node -e 'const p=require("./package.json"),fs=require("fs");let ok=0,miss=[];for(const[k,v]of Object.entries(p.exports))for(const f of(typeof v==="string"?[v]:Object.values(v)))fs.existsSync(f)?ok++:miss.push(k+" -> "+f);console.log(JSON.stringify({entries:Object.keys(p.exports),ok,missing:miss}))')` prints 9 entries (`.`, `./browser`, `./conversation`, `./files`, `./settings`, `./sidebar`, `./terminal`, `./i18n`, `./package.json`), `ok` 25 and `missing` `[]`. `grep -ohE 'AgentServerUIProviders|AgentServerUIRoot|CloudOrganizationBoundary|configureTelemetry|setTelemetryConsent' $L/dist/lib/index.d.ts | sort -u` lists all five. Remove the copy with `rm -rf $L`.
+- **Library build (`F26.lib-build`).** Never build inside the shared checkout (it regenerates i18n and typegen files the other runs hash); build in a copy: `L=$OH_VERIFY_RUN/private/f26-lib; mkdir -p $L; tar --exclude=./node_modules --exclude=./build --exclude=./.git --exclude=./dist -cf - . | tar -xf - -C $L; ln -s "$PWD/node_modules" $L/node_modules; (cd $L && npm run build:lib); echo "exit=$?"`. It ends with `✓ built in` and `exit=0` (about 30 s). Then `(cd $L && node -e 'const p=require("./package.json"),fs=require("fs");let ok=0,miss=[];for(const[k,v]of Object.entries(p.exports))for(const f of(typeof v==="string"?[v]:Object.values(v)))fs.existsSync(f)?ok++:miss.push(k+" -> "+f);console.log(JSON.stringify({entries:Object.keys(p.exports),ok,missing:miss}))')` prints 9 entries (`.`, `./browser`, `./conversation`, `./files`, `./settings`, `./sidebar`, `./terminal`, `./i18n`, `./package.json`), `ok` 25 and `missing` `[]`. `grep -ohE 'AgentServerUIProviders|AgentServerUIRoot|CloudOrganizationBoundary|configureTelemetry|setTelemetryConsent' $L/dist/lib/index.d.ts | sort -u` lists all five. Remove the copy with `rm -rf $L` (after the host-app bullet below when it runs).
- **Style scope (`F26.lib-style-scope`).** On your run's home page (`control-openhands browser goto /`) run `control-openhands browser eval "(() => { let total=0, unscoped=[]; for (const sh of document.styleSheets) { let rules; try { rules = sh.cssRules } catch { continue } const walk = (rs) => { for (const r of rs) { if (r.selectorText) { total++; if (!r.selectorText.includes('data-agent-server-ui')) unscoped.push(r.selectorText) } else if (r.cssRules) walk(r.cssRules) } }; walk(rules) } return { scopeRoots: document.querySelectorAll('[data-agent-server-ui]').length, total, unscopedCount: unscoped.length, sample: unscoped.slice(0, 12) } })()"`. `total` is in the thousands and `unscopedCount` is `2`: only `.go<hash>` and `.go<hash> > *`, react-hot-toast's runtime classes. `control-openhands browser eval "getComputedStyle(document.querySelector('[data-agent-server-ui]')).getPropertyValue('--oh-color-base').trim()"` is `#181818`. This proves the bundle is scoped, not that a host page is unaffected (`F26.lib-host-app`).
+- **Host app (`F26.lib-host-app`).** After the library build above (keep `$L` until this bullet is done), copy the checked-in [library host fixture](../../scripts/fixtures/library-host/README.md) (`.agents/skills/verify-openhands/scripts/fixtures/library-host/`) to a scratch directory outside the checkout, point its `@openhands/agent-canvas` dependency at `file:$L`, `npm install --ignore-scripts` there and start `QA_CANVAS_PACKAGE_DIR=$L npx vite --host 127.0.0.1 --port <owned-port>` in the background; write `http://127.0.0.1:<owned-port>/` to `$OH_VERIFY_RUN/workspace/library-host-url`. Mount: run `control-openhands browser goto "$(cat "$OH_VERIFY_RUN/workspace/library-host-url")" --allow-external`, `control-openhands browser wait '#outside'`, `control-openhands browser wait '#inside, #module-error:not(:empty)' --timeout 30000` and `control-openhands browser screenshot --feature F26.lib-host-app --name bare-package-host`. `control-openhands assert --field value --equals true -- browser eval '!!document.querySelector("#inside") && document.querySelector("#module-error").textContent === ""'` holds: `#inside` mounts inside `AgentServerUIRoot` with no module error. Then `control-openhands browser click '#inside'`, and `control-openhands assert --field text --equals 'Inside Canvas 1' -- browser text '#inside'` holds. Styles: link `$L/dist/assets/agent-canvas-*.css` from the host page, add an `AgentServerUIProviders` with `analytics: false` and `styleOverrides: {"--oh-color-base": "#101820"}` around `EmptyBrowserMessage`, reload and read `getComputedStyle` of the component's outer `flex flex-col … p-10` div and of a host element (`control-openhands browser eval`, `browser screenshot --feature F26.lib-host-app --name host-component`). Expected: the component is laid out as a flex column with Canvas colours, the override changes its base colour, and host elements keep browser defaults. Today (Known failure above) the component computes `display: block`, `padding: 0px`; the scope root carries `--oh-color-base: #101820` but no loaded rule uses it. Stop the Vite process and remove the scratch host and `$L` afterwards.
- **Docker conversation isolation (`F26.docker-conversation-runtime`).** Independently of the all-in-one Canvas deployment image, configure an owned source stack's Agent Server with `OH_CONVERSATION_RUNTIME=docker` and `OH_CONVERSATION_IMAGE` pointing to a compatible pinned Agent Server image. In SDK 1.55.1 the server creates a separate container with conversation, persistence and workspace mounts. A matching `-minimal` image can cover terminal/workspace isolation, but does not supply browser capabilities. Launch a confined conversation from Canvas, verify an actual model TerminalAction writes/reads a harmless marker under `/workspace`, and independently match the container hostname, mounted file and host-root absence. DELETE the conversation and verify its container and owned workspace disappear; restore original runtime settings. The 2026-10-17 test passed with the official 1.55.1 minimal image plus a small task-owned cloud-proxy/public-CA layer (no credentials). The unmodified image's model egress failed in this proxy-dependent environment; preserve that failure separately. This does not pass all-in-one packaging, browser or ACP recipes. Never print container/session keys or reuse a shared container as a fixture.
- **Docker and Helm prerequisites (`F26.docker-image`, `F26.docker-conversation-runtime`, `F26.helm-chart`).** If `docker info` cannot reach the daemon but `dockerd` is installed, start it (`setsid dockerd`); without one, record both Docker rows `blocked` naming the missing daemon. With a daemon, run the image with `-p 127.0.0.1:8000:8000`, open `http://localhost:8000/canvas` and expect the **Add a backend** step (`testid=onboarding-step-check-backend`, Next disabled until a key is typed); `onboarding-skip` leads to `testid=api-key-entry-screen`. With `-e AGENT_CANVAS_ALLOW_LAN_SESSION_KEY=true` the key is injected and first run opens on Choose your agent, under the telemetry consent dialog. `npm run build:docker` behind a TLS-intercepting proxy needs the proxy CA inside the build stages (without it `uv pip install` / `pip install` in the final stage fail with `CERTIFICATE_VERIFY_FAILED: self-signed certificate in certificate chain`), and Docker Hub may answer 429. Build a scratch copy of the Dockerfile that only adds the CA, never an edited `docker/Dockerfile`: after `FROM node:24-slim AS frontend-build` add `COPY --from=ccrca ca-bundle.crt /tmp/qa-ca.crt` and `ENV NODE_EXTRA_CA_CERTS=/tmp/qa-ca.crt npm_config_cafile=/tmp/qa-ca.crt`, after the first `USER root` add the same `COPY` and `ENV SSL_CERT_FILE=/tmp/qa-ca.crt PIP_CERT=/tmp/qa-ca.crt REQUESTS_CA_BUNDLE=/tmp/qa-ca.crt UV_NATIVE_TLS=1`, then `docker build -f <copy> --build-context ccrca=<dir holding ca-bundle.crt> --network host --build-arg AGENT_SERVER_IMAGE=<images.agentServer>:<versions.agentServer>-python --build-arg AGENT_SERVER_VERSION=<versions.agentServer> --build-arg AUTOMATION_VERSION=<versions.automation> --build-arg VITE_BASE_PATH=/canvas -t qa-agent-canvas:target .` (the pins `scripts/docker-build.mjs` reads from `config/defaults.json`; about 5 min on a warm cache). Publish it on a free port (`-p 127.0.0.1:<free>:8000`; 8000 and the 188xx blocks may belong to other runs). Docker Hub may answer `429 Too Many Requests` for `node:24-slim`; in the scratch copy only, replace both `FROM node:24-slim` lines with `FROM mirror.gcr.io/library/node:24-slim` (arrange; say so in the note). Run the unmodified image with `docker run -d --name <qa-name> -p 127.0.0.1:<free>:8000 qa-agent-canvas:target`. On a kernel without IPv6, `docker logs <qa-name>` shows `IPv6 is not available; binding the frontend to 0.0.0.0.` and `Static-server + proxy listening on http://0.0.0.0:8000/canvas/`, `/canvas/` answers `200` within about 25 s and `docker inspect -f '{{.State.Status}}' <qa-name>` is still `running` minutes later (#18178 fixed by #18180, verified 2026-10-13); `docker exec <qa-name> ls /opt/agent-canvas` lists the `mars-*.mjs` bridge modules (#18343). The container's generated key is `docker exec <qa-name> cat /home/openhands/.openhands/agent-canvas/api-key.txt` (write it to `$OH_VERIFY_RUN/private/`, fill with `--value-file`). Settings → Application (`/canvas/settings/app`) shows the image's `Version` (`agent-canvas-update-card`), which matches `jq -r .version package.json`. Remove the containers and the image afterwards (`docker rm -f`, `docker rmi`). For `F26.lib-host-app`, arrange a normal separate browser host with matching declared peers and the built package; the 2026-10-17 real host reproduced #126 above. A missing old example is not itself a runtime prerequisite. Install a compatible task-local Helm if the binary is missing, then run `helm lint helm/agent-canvas` and `helm template qa helm/agent-canvas`: the defaults render the StatefulSet, Services and ServiceAccount; Ingress is disabled by default and needs a separate `--set ingress.enabled=true` template check. The 2026-10-17 Helm 3.19.0 static checks passed in both variants. Live `helm install` follows `helm/agent-canvas/README.md` and requires an authorized disposable Kubernetes context; absent kubeconfig/in-cluster credentials are a concrete live prerequisite, not a missing-binary excuse.
- **Desktop splash and window (`F26.desktop-boot-splash`, `F26.desktop-main-window`).** Start a private display and the app in one process group: `Q=$OH_VERIFY_RUN/private/f26-desktop; mkdir -p $Q/shots; setsid nohup sh -c "echo \$\$ > $Q/pgid; Xvfb :78 -screen 0 1440x1000x24 -nolisten tcp & sleep 1; DISPLAY=:78 HOME=$Q/home OH_CANVAS_SAFE_STATE_DIR=$Q/state LOCAL_BACKEND_API_KEY=qa-f26-desktop-key OH_SECRET_KEY=qa-f26-secret DO_NOT_TRACK=1 exec node_modules/.bin/electron --no-sandbox electron" > $Q/desktop.log 2>&1 &`, then capture frames: `for i in $(seq 1 40); do sleep 1; DISPLAY=:78 import -window root $Q/shots/s$(printf %02d $i).png; done`. Read the frames: the first seconds show the splash (logo, `OpenHands Agent Canvas`, `AI coding agent interface`, spinner, `Starting backend services…`, then lines such as `agent-server: {"asctime": ...`, the note `First launch downloads Python + the OpenHands agent server.` and a **Show details** button); a later frame shows the main window with a `File Edit View Window` menu and the first-run **Choose your agent** step under the consent dialog naming `the local backend "Local" at http://localhost:8000` (about 30 s on a warm cache, minutes on the first download). While it runs, `curl -s http://127.0.0.1:8000/mars/health | head -c 15` prints `<!DOCTYPE html>`: the desktop stack's ingress hosts no bridge (`marsWebBridge: false`, `F26.mars-web-bridge`). Copy the frames you cite into `$OH_VERIFY_RUN/evidence/F26.desktop-boot-splash/` and `.../F26.desktop-main-window/`. Stop in two steps: `kill -TERM $(cat $Q/pgid)` signals only the Electron launcher (it quits and stops its backend services); after about 10 s ports 8000, 18000, 18001 and 3001 are free again. Then `kill -TERM -- -$(cat $Q/pgid)` ends Xvfb; `ps -o pid,cmd -g $(cat $Q/pgid)` lists nothing.
diff --git a/.agents/skills/verify-openhands/references/feature-map/F28-automation-setup-page.md b/.agents/skills/verify-openhands/references/feature-map/F28-automation-setup-page.md
index 1185805090..1092d722df 100644
--- a/.agents/skills/verify-openhands/references/feature-map/F28-automation-setup-page.md
+++ b/.agents/skills/verify-openhands/references/feature-map/F28-automation-setup-page.md
@@ -28,7 +28,7 @@ Source: `src/routes/automation-setup-new.tsx`, `src/components/features/automati
- `F28.event-trigger`: choosing **Event** replaces the schedule with Event source (default **GitHub**, sent as `github`; its menu lists the webhook sources registered on this backend plus **Custom**), Event key (default `issue_comment.created`), **Add Filter** (one free-text filter input), **Ask Agent**, the notice `Event sources need a public URL to deliver webhooks to this instance.`, and a **Test event payload** pre-filled with a JSON object for that source and key. A source the backend has not registered (menu **Custom**, then a typed name) adds **Register this as a custom webhook source**, checked by default.
- `F28.timeout`: **Add Timeout** adds an optional **Timeout seconds** number field with a remove button; `1800` saves and the draft reads **Ready to test**.
- `F28.draft-save`: **Save draft** works on any form, valid or not; the first save replaces the URL with `?draftId=<id>`, shows the **Draft saved** bar (relative time, **View test runs**, **Test**) and the toast `Draft saved in this conversation.`, and the save state reads **Saved just now**. The draft appears in the dashboard Drafts group (F21).
-- `F28.draft-validation`: the server validates on save, not on blur. An invalid draft is still saved (`dispatchable: false`) and its first error replaces **Ready to test** in the draft bar: an empty prompt shows `Field required` (also under Prompt), a bad custom cron shows `Value error, Invalid cron expression: every day`. A field error stays until the next save. **Test** stays enabled on a saved invalid draft: it saves again, opens **Test runs** with the same error in a red alert and dispatches nothing. Known failure (reproduced twice 2026-10-17, [#18601](https://github.com/OpenHands/OpenHands/issues/18601)): after a successful draft Test, saving an invalid cron and reloading shows a red `Latest test passed` pill instead of the current validation error, although the API marks the draft non-dispatchable. The Test runs alert still shows the correct error.
+- `F28.draft-validation`: the server validates on save, not on blur. An invalid draft is still saved (`dispatchable: false`) and its first error replaces **Ready to test** in the draft bar: an empty prompt shows `Field required` (also under Prompt), a bad custom cron shows `Value error, Invalid cron expression: every day`. A field error stays until the next save. **Test** stays enabled on a saved invalid draft: it saves again, opens **Test runs** with the same error in a red alert and dispatches nothing. The current validation error comes before any test status: after a successful draft Test the pill reads **Latest test passed**, and once an invalid cron is saved it reads the cron error (also after a reload, at desktop and phone widths) while the API marks the draft non-dispatchable.
- `F28.draft-save-error`: a draft the server refuses outright (a timeout above 1800 returns `422`) sets the save state to **Save failed** and shows one error toast with the reason the API gives (a timeout of 1801 reads `Value error, timeout must not exceed 1800 seconds (30 minutes)`). The draft bar drops its validity pill (no stale **Ready to test**) until a save succeeds.
- `F28.required-before-test`: Test and Create check locally first and send nothing when a required field is empty: `Add a prompt before continuing.` (Prompt or Plugin kind), `Add a plugin source before continuing.`, `Add Python code before continuing.`, `Add an entrypoint command before continuing.`, `Add a setup script path before continuing.`, `Add a setup script before continuing.`; an event Test with a payload that is not a JSON object shows `Enter a valid JSON object before testing this event automation.` A blank Name is not an error: it is derived from the prompt.
- `F28.test-gated`: in create mode **Test** stays disabled until the draft is saved and unchanged since; editing any field after a save disables it again until the next **Save draft**.
@@ -78,7 +78,9 @@ control-openhands browser screenshot --feature F28.schedule-presets --name weekl
control-openhands assert --field value --equals '30 10 * * 3' -- browser value 'testid=automation-setup-custom-schedule'
```
-```sh known-bug F28.draft-validation stale-test-success
+Validation error after a passed Test (`F28.draft-validation`):
+
+```sh
control-openhands browser viewport desktop
control-openhands browser goto /automations/setup
control-openhands browser fill 'testid=automation-setup-name' QA_Stale_Test_Success
@@ -107,8 +109,8 @@ control-openhands browser click 'testid=automation-setup-save-draft'
control-openhands browser hover 'testid=automation-setup-name'
control-openhands browser reload
control-openhands assert --field body.dispatchable --equals false -- api GET "/api/automation/v1/drafts/$QA_DRAFT_ID"
-control-openhands browser screenshot --feature F28.draft-validation --name stale-test-success
-control-openhands assert --field text --equals 'Value error, Invalid cron expression: every day' --bug-shows 'Latest test passed' -- browser text 'testid=automation-setup-draft-validity' # bug
+control-openhands browser screenshot --feature F28.draft-validation --name validation-after-test
+control-openhands assert --field text --equals 'Value error, Invalid cron expression: every day' -- browser text 'testid=automation-setup-draft-validity'
```
@@ -119,5 +121,5 @@ control-openhands assert --field text --equals 'Value error, Invalid cron expres
- Change one field per Save draft when checking a validation message: a save shows only the first server error, so two edits in one save hide the second.
- A cold load of `/automations/setup?draftId=<id>` or `?automationId=<id>` (a reload or a typed URL) opens the form: re-checked 2026-10-14 on a `control-openhands launch` of main with this family (both read their Name back). An earlier instance of unknown build stayed on **Loading saved draft…** for more than 20 s; if that returns, compare with opening the same draft from the dashboard.
- The draft bar duplicates the prompt error: `Field required` appears both under Prompt and in the bar.
-- Kebab **Edit** on a Custom Python automation shows the code editor instead of the saved bundle summary: open bug #18341 (`F23`), with fixes in PRs #18353 and #18444.
+- Kebab **Edit** on a Custom Python (tarball) automation shows the **Existing code bundle** summary, not the code editor, and Save keeps that bundle without asking for upload-time fields (`F23.edit-open`, `F23.edit-save`).
- Escape on the setup page does nothing, by design (`F23.edit-close`).
diff --git a/.agents/skills/verify-openhands/references/feature-map/README.md b/.agents/skills/verify-openhands/references/feature-map/README.md
index 7845d9fcd5..5660a6d9fd 100644
--- a/.agents/skills/verify-openhands/references/feature-map/README.md
+++ b/.agents/skills/verify-openhands/references/feature-map/README.md
@@ -5,7 +5,7 @@ of Agent Canvas. Read this index before driving the app, then open the matching
feature file and run its recipe with `control-openhands`. The map is not a claim
that every row passes today; the run's evidence ledger is.
-Maintenance baseline: main@045290739f0af2ae4cc89d2a209c6bd930bb81ee (2026-10-17). The next maintenance pass starts from this commit; a pass proposes the next baseline in its PR, and merging that PR accepts it.
+Maintenance baseline: main@79abd32d856d86a551898ba14ce03f686d453663 (2026-10-19). The next maintenance pass starts from this commit; a pass proposes the next baseline in its PR, and merging that PR accepts it.
## Baseline preconditions
@@ -22,6 +22,15 @@ Maintenance baseline: main@045290739f0af2ae4cc89d2a209c6bd930bb81ee (2026-10-17)
rate limit), `llm preset deepseek --no-validate` still saves the profiles:
recipes that only need a conversation to exist can run on, and every
recipe that reads a model reply is `blocked`, never a pass.
+- Known issue [OpenHands/software-agent-sdk#6088](https://github.com/OpenHands/software-agent-sdk/issues/6088)
+ (reproduced 2026-10-19): a fresh install of the pinned Agent Server 1.55.1
+ resolves litellm 1.106.0 next to mcp 1.30.0. From then on every model call
+ fails with `cannot import name 'CacheableResult' from 'mcp.types'`, and the
+ conversation ends in `error`. `doctor` stays `ok`. To run model-backed recipes
+ meanwhile, write `$OH_VERIFY_RUN/private/home/.config/uv/uv.toml` containing
+ `[exclude-newer-package]` and `litellm = "2026-10-18T00:00:00Z"`, then run
+ `control-openhands restart`. That run's Agent Server then installs litellm
+ 1.105.1.
- For critic checks, use an available, authorized OpenHands (`openhands/…`) or compatible OpenHands LLM-proxy profile, regardless of the main profile used for the pass. Its LLM key is reused when Critic API Key is empty. For a non-default proxy deployment, use its compatible critic endpoint. Restore the previous profile afterwards; see [F06](F06-agent-activity.md). Do not declare the critic blocked merely because the main model uses another provider. This does not authorize new credentials, additional budget, or sending private test data to another service.
- The browser is at the desktop viewport (1440×1000) unless a recipe says
otherwise; `control-openhands browser viewport phone` is 390×844.
diff --git a/.agents/skills/verify-openhands/scripts/fixtures/library-host/README.md b/.agents/skills/verify-openhands/scripts/fixtures/library-host/README.md
new file mode 100644
index 0000000000..ca594a52b3
--- /dev/null
+++ b/.agents/skills/verify-openhands/scripts/fixtures/library-host/README.md
@@ -0,0 +1,13 @@
+# Library host fixture
+
+A minimal Vite 8 browser host that imports `@openhands/agent-canvas` by bare
+package name, with React 19.3.0 peers, and renders a button inside
+`AgentServerUIRoot`. `#outside` is a host element outside the Canvas scope, and
+`#module-error` shows any native `window` error. There is no `require` shim and
+no mocked module response.
+
+Copy this directory to a scratch directory outside the checkout. Replace
+`REPLACE_WITH_LIBRARY_BUILD_DIR` in `package.json` with the checkout copy where
+`npm run build:lib` ran (`$L`), then run `npm install --ignore-scripts` and
+`QA_CANVAS_PACKAGE_DIR=$L npx vite --port <owned-port>`.
+[F26](../../../references/feature-map/F26-runtime-variants.md) (`F26.lib-host-app`) drives it.
diff --git a/.agents/skills/verify-openhands/scripts/fixtures/library-host/host.js b/.agents/skills/verify-openhands/scripts/fixtures/library-host/host.js
new file mode 100644
index 0000000000..c64230564f
--- /dev/null
+++ b/.agents/skills/verify-openhands/scripts/fixtures/library-host/host.js
@@ -0,0 +1,8 @@
+import React from "react";
+import { createRoot } from "react-dom/client";
+import { AgentServerUIRoot } from "@openhands/agent-canvas";
+function App() {
+ const [count, setCount] = React.useState(0);
+ return React.createElement(AgentServerUIRoot, {}, React.createElement("button", { id: "inside", onClick: () => setCount((value) => value + 1) }, "Inside Canvas " + count));
+}
+createRoot(document.getElementById("root")).render(React.createElement(App));
diff --git a/.agents/skills/verify-openhands/scripts/fixtures/library-host/index.html b/.agents/skills/verify-openhands/scripts/fixtures/library-host/index.html
new file mode 100644
index 0000000000..cdba9cf7c2
--- /dev/null
+++ b/.agents/skills/verify-openhands/scripts/fixtures/library-host/index.html
@@ -0,0 +1 @@
+<!doctype html><html lang="en"><head><meta charset="UTF-8"><title>Canvas package consumer</title></head><body><h1 id="outside">Normal Vite host — outside Canvas</h1><pre id="module-error" role="alert"></pre><div id="root"></div><script>window.addEventListener("error",(event)=>{document.getElementById("module-error").textContent=event.message;});</script><script type="module" src="/host.js"></script></body></html>
diff --git a/.agents/skills/verify-openhands/scripts/fixtures/library-host/package.json b/.agents/skills/verify-openhands/scripts/fixtures/library-host/package.json
new file mode 100644
index 0000000000..37d9f2581a
--- /dev/null
+++ b/.agents/skills/verify-openhands/scripts/fixtures/library-host/package.json
@@ -0,0 +1,12 @@
+{
+ "name": "canvas-library-browser-proof",
+ "private": true,
+ "type": "module",
+ "dependencies": {
+ "@openhands/agent-canvas": "file:REPLACE_WITH_LIBRARY_BUILD_DIR",
+ "react": "19.3.0",
+ "react-dom": "19.3.0",
+ "react-router": "7.18.2",
+ "vite": "8.0.16"
+ }
+}
diff --git a/.agents/skills/verify-openhands/scripts/fixtures/library-host/vite.config.mjs b/.agents/skills/verify-openhands/scripts/fixtures/library-host/vite.config.mjs
new file mode 100644
index 0000000000..550d42536c
--- /dev/null
+++ b/.agents/skills/verify-openhands/scripts/fixtures/library-host/vite.config.mjs
@@ -0,0 +1,10 @@
+import { defineConfig } from "vite";
+import { resolve } from "node:path";
+import { realpathSync } from "node:fs";
+if (!process.env.QA_CANVAS_PACKAGE_DIR) throw new Error("Set QA_CANVAS_PACKAGE_DIR to the built library checkout");
+const packageDirectory = resolve(process.env.QA_CANVAS_PACKAGE_DIR);
+export default defineConfig({
+ server: { host: "127.0.0.1", strictPort: true, fs: { allow: [process.cwd(), packageDirectory, realpathSync(resolve(packageDirectory, "node_modules"))] } },
+ optimizeDeps: { include: ["@openhands/agent-canvas"] },
+ resolve: { dedupe: ["react", "react-dom", "react-router"] },
+});
diff --git a/.pr/bugs/2026-10-19/git-chip-icons-light-themes/default-desktop.png b/.pr/bugs/2026-10-19/git-chip-icons-light-themes/default-desktop.png
new file mode 100644
index 0000000000..9ec1057d70
Binary files /dev/null and b/.pr/bugs/2026-10-19/git-chip-icons-light-themes/default-desktop.png differ
diff --git a/.pr/bugs/2026-10-19/git-chip-icons-light-themes/lightplus-desktop.png b/.pr/bugs/2026-10-19/git-chip-icons-light-themes/lightplus-desktop.png
new file mode 100644
index 0000000000..a5fcb556f2
Binary files /dev/null and b/.pr/bugs/2026-10-19/git-chip-icons-light-themes/lightplus-desktop.png differ
diff --git a/.pr/bugs/2026-10-19/git-chip-icons-light-themes/lightplus-phone.png b/.pr/bugs/2026-10-19/git-chip-icons-light-themes/lightplus-phone.png
new file mode 100644
index 0000000000..4b3bb40bb1
Binary files /dev/null and b/.pr/bugs/2026-10-19/git-chip-icons-light-themes/lightplus-phone.png differ
diff --git a/.pr/bugs/2026-10-19/git-chip-icons-light-themes/solarized-desktop.png b/.pr/bugs/2026-10-19/git-chip-icons-light-themes/solarized-desktop.png
new file mode 100644
index 0000000000..38889f7e2f
Binary files /dev/null and b/.pr/bugs/2026-10-19/git-chip-icons-light-themes/solarized-desktop.png differ
diff --git a/.pr/bugs/2026-10-19/git-chip-icons-light-themes/solarized-phone.png b/.pr/bugs/2026-10-19/git-chip-icons-light-themes/solarized-phone.png
new file mode 100644
index 0000000000..614ba87930
Binary files /dev/null and b/.pr/bugs/2026-10-19/git-chip-icons-light-themes/solarized-phone.png differ
diff --git a/.pr/bugs/2026-10-19/library-no-stylesheet/control-with-app-css.png b/.pr/bugs/2026-10-19/library-no-stylesheet/control-with-app-css.png
new file mode 100644
index 0000000000..664e2dedf1
Binary files /dev/null and b/.pr/bugs/2026-10-19/library-no-stylesheet/control-with-app-css.png differ
diff --git a/.pr/bugs/2026-10-19/library-no-stylesheet/unstyled-desktop.png b/.pr/bugs/2026-10-19/library-no-stylesheet/unstyled-desktop.png
new file mode 100644
index 0000000000..8f061052c4
Binary files /dev/null and b/.pr/bugs/2026-10-19/library-no-stylesheet/unstyled-desktop.png differ
diff --git a/.pr/bugs/2026-10-19/library-no-stylesheet/unstyled-phone.png b/.pr/bugs/2026-10-19/library-no-stylesheet/unstyled-phone.png
new file mode 100644
index 0000000000..63807be8f5
Binary files /dev/null and b/.pr/bugs/2026-10-19/library-no-stylesheet/unstyled-phone.png differ
diff --git a/.pr/map-maintenance/2026-10-19/F02.error-toasts-light-plus.png b/.pr/map-maintenance/2026-10-19/F02.error-toasts-light-plus.png
new file mode 100644
index 0000000000..992000035e
Binary files /dev/null and b/.pr/map-maintenance/2026-10-19/F02.error-toasts-light-plus.png differ
diff --git a/.pr/map-maintenance/2026-10-19/F02.mobile-drawer-modal-escape.png b/.pr/map-maintenance/2026-10-19/F02.mobile-drawer-modal-escape.png
new file mode 100644
index 0000000000..43bf492b88
Binary files /dev/null and b/.pr/map-maintenance/2026-10-19/F02.mobile-drawer-modal-escape.png differ
diff --git a/.pr/map-maintenance/2026-10-19/F06.error-banner-light-plus.png b/.pr/map-maintenance/2026-10-19/F06.error-banner-light-plus.png
new file mode 100644
index 0000000000..1f6cc2d24f
Binary files /dev/null and b/.pr/map-maintenance/2026-10-19/F06.error-banner-light-plus.png differ
diff --git a/.pr/map-maintenance/2026-10-19/F06.hook-events-light-plus.png b/.pr/map-maintenance/2026-10-19/F06.hook-events-light-plus.png
new file mode 100644
index 0000000000..66e01b8cb2
Binary files /dev/null and b/.pr/map-maintenance/2026-10-19/F06.hook-events-light-plus.png differ
diff --git a/.pr/map-maintenance/2026-10-19/F08.overview-diffs-light-plus.png b/.pr/map-maintenance/2026-10-19/F08.overview-diffs-light-plus.png
new file mode 100644
index 0000000000..5701dfd6de
Binary files /dev/null and b/.pr/map-maintenance/2026-10-19/F08.overview-diffs-light-plus.png differ
diff --git a/.pr/map-maintenance/2026-10-19/F16.language-rtl-switches.png b/.pr/map-maintenance/2026-10-19/F16.language-rtl-switches.png
new file mode 100644
index 0000000000..e0dc71b7aa
Binary files /dev/null and b/.pr/map-maintenance/2026-10-19/F16.language-rtl-switches.png differ
diff --git a/.pr/map-maintenance/2026-10-19/F19.launch-parameters-desktop.png b/.pr/map-maintenance/2026-10-19/F19.launch-parameters-desktop.png
new file mode 100644
index 0000000000..38e0f61ddd
Binary files /dev/null and b/.pr/map-maintenance/2026-10-19/F19.launch-parameters-desktop.png differ
diff --git a/.pr/map-maintenance/2026-10-19/F19.launch-parameters-phone.png b/.pr/map-maintenance/2026-10-19/F19.launch-parameters-phone.png
new file mode 100644
index 0000000000..4e922d7bec
Binary files /dev/null and b/.pr/map-maintenance/2026-10-19/F19.launch-parameters-phone.png differ
diff --git a/.pr/map-maintenance/2026-10-19/F23.edit-save-bundle.png b/.pr/map-maintenance/2026-10-19/F23.edit-save-bundle.png
new file mode 100644
index 0000000000..442f223211
Binary files /dev/null and b/.pr/map-maintenance/2026-10-19/F23.edit-save-bundle.png differ
diff --git a/.pr/map-maintenance/2026-10-19/F23.phone-long-name-ltr.png b/.pr/map-maintenance/2026-10-19/F23.phone-long-name-ltr.png
new file mode 100644
index 0000000000..d0a0d988ed
Binary files /dev/null and b/.pr/map-maintenance/2026-10-19/F23.phone-long-name-ltr.png differ
diff --git a/.pr/map-maintenance/2026-10-19/F23.phone-rtl-header.png b/.pr/map-maintenance/2026-10-19/F23.phone-rtl-header.png
new file mode 100644
index 0000000000..bdcea907a1
Binary files /dev/null and b/.pr/map-maintenance/2026-10-19/F23.phone-rtl-header.png differ
diff --git a/.pr/map-maintenance/2026-10-19/F25.edit-escape-phone.png b/.pr/map-maintenance/2026-10-19/F25.edit-escape-phone.png
new file mode 100644
index 0000000000..2a09e93ca5
Binary files /dev/null and b/.pr/map-maintenance/2026-10-19/F25.edit-escape-phone.png differ
diff --git a/.pr/map-maintenance/2026-10-19/F25.manage-add-phone-escape.png b/.pr/map-maintenance/2026-10-19/F25.manage-add-phone-escape.png
new file mode 100644
index 0000000000..cd95129ef7
Binary files /dev/null and b/.pr/map-maintenance/2026-10-19/F25.manage-add-phone-escape.png differ
diff --git a/.pr/map-maintenance/2026-10-19/F26.lib-host-app-mounted.png b/.pr/map-maintenance/2026-10-19/F26.lib-host-app-mounted.png
new file mode 100644
index 0000000000..9ad09d7200
Binary files /dev/null and b/.pr/map-maintenance/2026-10-19/F26.lib-host-app-mounted.png differ
diff --git a/.pr/map-maintenance/2026-10-19/F28.draft-validation-error.png b/.pr/map-maintenance/2026-10-19/F28.draft-validation-error.png
new file mode 100644
index 0000000000..34f78b89d7
Binary files /dev/null and b/.pr/map-maintenance/2026-10-19/F28.draft-validation-error.png differ