Automate browser based workflows with AI
Python aiapiautomationbrowserbrowser-automationcomputergptllmplaywrightpowerautomatepuppeteerpythonrpaseleniumvisionworkflow
Remove internal working files from docs/ (#8878)
nickjtch-skyvern624cc733130f
diff --git a/docs/changelog-emails/2026-05-may.md b/docs/changelog-emails/2026-05-may.mddeleted file mode 100644index ff7953299..000000000--- a/docs/changelog-emails/2026-05-may.md+++ /dev/null@@ -1,201 +0,0 @@---# Skyvern Changelog — April & May 2026--*Two months of shipping in one email — a draggable editor, cleaner MCP OAuth, stronger run controls, and plenty more. Headliners first, the full list at the end.*-------## April 2026--### Workflow Copilot v2--The workflow copilot got a lot more hands-on. **Hard-cancel** a run mid-flight instead of waiting it out, and **follow it live** as it **streams status narration** and surfaces **block-level events** (starts, completions, errors) while your workflow runs. When a test trips, it hands back the **work-in-progress workflow** so you can grab the draft and keep building. Self-healed retries are now visually separated from real failures, and a quick `/discover` in chat passes the whole build straight to the copilot.---------### MCP OAuth, Across Clients--Connecting Skyvern over MCP is now a clean **OAuth** flow. **Multi-org users** get an organization picker right in the authorize step. Skyvern runs as a native **OpenAI Apps SDK** app, ships **safety hints** for the ChatGPT connector, documents **Codex** remote OAuth end-to-end, and adds **OpenClaw** as a new target (`skyvern setup openclaw`). Under the hood: production OAuth callbacks, plus per-tool latency and error telemetry.---------### `skyvern schedule` + `skyvern config` CLI--Your terminal is now a control panel. **`skyvern schedule`** creates, lists, toggles, and deletes workflow schedules without leaving the command line. **`skyvern config`** reads and writes org-level settings — timeouts, concurrency, default proxy — the same way. Both run over the MCP protocol, so Claude Code, Codex, and other MCP clients can use them through the same interface.---------### AI Summarize for Workflow Outputs--Stop squinting at nested JSON. A new **Summarize with AI** button sits next to every block and workflow output — one click turns raw data into a plain-language summary of what your workflow extracted.---------### Workflow-Level Error Code Mapping--Set your error handling once — not on every block. Workflows now take an `error_code_mapping` field that **every block inherits automatically**. Define the defaults at the top, and override per-block only where you actually need to.---------### Debugger Remembers Last-Used Values--Less busywork between test runs. The debugger now **pre-fills the Run dialog with the values from your last run**, so you're not re-typing the same inputs every time.---------## May 2026--### Draggable Workflow Editor--The editor is now **drag-and-drop** — reorder your whole workflow, not just tweak blocks in place. Grab any block by its handle and **drop it where it belongs**: top level, inside a branch, deep in a loop. **Collapse** blocks and entire nested containers to focus, and your layout is **remembered per workflow** across sessions. It's fully **keyboard-driven** with screen-reader announcements, and every block's config **autosaves as you type**. (Recording mode still freezes drag, so a mid-record reorder can't shift block identities under the recorder.)---------### Redesigned Run Page--We redesigned the run page around a compact **timeline sidebar** and a dedicated **detail panel** — scan every block at a glance, then open any one for its inputs, output, and diagnostics. A live **N-of-M completion counter** tracks progress and visibly flags blocks that never ran. The **app sidebar** got a refresh in the same pass, for cleaner navigation everywhere.---------### While Loop Block--Loop until you're done — not a fixed number of times. The new **While Loop block** repeats a sub-workflow **until your condition is met**: paginate until there's no Next button, retry until a status flips, drain a queue until it's empty.---------### Analytics Dashboard--Your automations, by the numbers. The new **`/analytics` page** breaks down **run volume, outcomes, and trends over time** in one view — no exports, no spreadsheets. (Gated per-org, so it lights up only where it's enabled.)---------### Workflow Copilot, Steadier--The copilot is steadier when sessions get long or messy. A persistent **turn-narrative bubble** tracks what the agent is doing — and **survives page reloads**, so you never lose your place. Hit an unexpected error? It's **recoverable now**: instead of a frozen chat, you get a clear explanation and a **specific follow-up question** drawn from what it actually found missing. The **diff view** finally catches changes **nested inside conditionals and loops**, and we cleared a batch of edge cases — credential scope leaking after an edit, session context lost across turns, accepted proposals getting dropped.---------### More Control Over Your Runs--Manage runs in bulk — cancel, retry, cap. The new **Retry API** re-runs a workflow **from the start or from any block** — no rebuilding inputs. **`POST /runs/cancel`** cancels **many runs in a single request** and tells you exactly which succeeded and which didn't. And two new guardrails — a **runtime limit** and a **per-workflow step cap** — stop runaway executions before they burn resources.---------### Self-Hosting: Custom Proxies & CDP Auth Headers--Route any run through **your own proxy** — just pass `{"proxy_location": {"url": "http://user:pass@host:port"}}` on any task, workflow, or browser-session request; no managed proxy infrastructure required. And **CDP connect headers** are now a first-class workflow setting — your provider's auth headers are masked, stored, and applied to the **CDP connection only**, never forwarded to target sites. (Plus: msedge is the new default browser for self-hosted templates, and the Docker Compose quickstart bootstraps your `.env` files on first run.)---------### Browser Profiles, End to End--Saved browser profiles — the logged-in sessions your workflows reuse — are easier to set up and manage. A **guided, step-by-step setup** walks you through creating one, a dedicated **management page** lets you **search, rename, and organize**, and saving now gives you **instant feedback** with an in-progress row the moment you hit save.---------## Quick List — April & May--**New features**-- **Credits per run & run-level cost breakdown** — a Credits badge on run pages, and task records now track LLM, proxy, and captcha costs as separate fields via the API-- **Credential Folders** — organize credentials into named folders, with a "Move to folder" control and a folder filter that persists across credential tabs-- **Multiple schedules per workflow** — independent schedules on one workflow (e.g. weekdays at 9am *and* weekends at noon)-- **Google Sheets connector** (open source) — the Sheets auth flow is now part of the OSS package-- **Tagging API** — tag CRUD, tag history, tag-key management, and batch tag endpoints-- **Blank-canvas workflow entry point** — a "Blank Workflow" option plus an "Or start from a description" handoff to Discover-- **MCP setup in cloud settings** — a dedicated MCP configuration section in account settings-- **Action Block selector + AI fallback fields** — direct control over element targeting and fallback behavior-- **Block-scoped parameter prompts in the debugger** — prompts scoped to the specific block under test-- **Runs tab on the browser session page** — every run tied to a given session, in one place-- **Smarter element targeting** — deterministic HTML element-tree compression for more reliable selectors-- **Saudi Arabia proxy** — `RESIDENTIAL_SA` added as a geolocation option-- **Password manager migration notice** — UI banner when password manager settings need updating-- **Loop block failure-handling UI** — clearer display of how failures are handled within loop blocks-- **OpenAI CUA model** now configurable via feature flag--**Improvements**-- **Cross-run extraction cache** — recurring scheduled workflows reuse extraction results across runs when page content hasn't changed, cutting redundant LLM calls-- MCP runs are now classified as `manual`, matching UI-triggered model routing and queueing-- MCP tool call performance telemetry tracks latency and errors across all MCP tool invocations-- Generic remote browser vendor support — evaluate third-party CDP providers via a JSON config, including HTTP/HTTPS DevTools endpoints-- Scheduled-trigger badge on the global Runs page-- Faster GET workflow run API-- Faster script-mode runs (skip speculative extract-action steps) and faster proxy selection in script mode-- Reduced scroll-into-view settle time for snappier agent actions-- LLM router retries with a fallback model when a response is cut off by a length limit-- Prompt token cap prevents context-window overflows on high-complexity workflows-- Extraction prompts capped in size before LLM calls, reducing `429 RESOURCE_EXHAUSTED` errors on high-volume workflows-- Hardened prompt input sanitization — untrusted page content is sanitized consistently at the template layer-- Automatic webhook retry on transient infrastructure failures-- Signed artifact content URLs with configurable expiry, plus presigned-URL fallback for self-hosted downloads-- Archived-artifact status indicator on runs-- Output-parameter size cap to avoid oversized payloads-- Double-click action support and PDF embed detection inside multi-frame pages-- Content blocking disabled for authenticated browser launches, improving login success rates-- Browser dialogs (alert/confirm/prompt) surfaced to the agent so it can adapt-- TOTP scoped to the active credential, and shown as placeholder text instead of pre-filled-- Clearer Browsers vs. Browser Profiles descriptions, and a toast action-button polish pass-- msedge is the default browser for self-hosted templates; quickstart bootstraps `.env` files-- Fewer screenshots captured for API-triggered runs, lowering storage overhead-- Auto-route browser type from browser profile source-- Reduced scrape-phase mouse movement on recaptcha-protected sites-- Renamed browser profile reset endpoint to `/browser_session/reset_profile` for clarity-- Removed deprecated Anthropic and Bedrock Claude model configs--**Bug fixes**-- Fixed workflow run download links showing a storage path instead of the real filename-- Fixed the Schedules panel not being dismissable from the workflow canvas-- Fixed scheduled runs not appearing on the global Runs page-- Fixed webhook delivery silently failing on transient 403, 5xx, or network errors-- Fixed webhook delivery failing on URLs with leading/trailing spaces-- Fixed popup pages and XHR-triggered downloads not being captured in runs and recordings-- Fixed child tasks not terminating and runs getting resurrected after a cleanup-cron timeout-- Fixed reCAPTCHA v2 invisible variant being misidentified as v3-- Fixed inline Press & Hold captchas not solving when the challenge renders without a modal wrapper-- Fixed the large `+` button in the editor closing the block menu instead of pinning it open-- Fixed the browser profile dropdown in Start-node settings clipping its last item-- Fixed credential tests started from the UI running on a lower-priority tier-- Fixed several Copilot edge cases — credential scope after edits, session context across turns, dropped proposals, and feasibility stalls-- Fixed copilot proposing a workflow it had just failed to build-- Fixed embedded multi-placeholder secrets not resolving in workflow credentials-- Fixed self-serve billing for accounts on top-up credit plans-- Fixed 2FA entry on pages with custom overlay input components-- Fixed CSV parsing failing on wide-format files with large header rows-- Fixed browser crash on sites with cross-origin stylesheets (MUI-based apps)-- Fixed context-window overflow failures being misdiagnosed as generic errors-- Fixed workflow file-download blocks losing `download_suffix`-- Fixed cached scripts generating incorrect `fill_form()` calls-- Fixed run recordings missing end timestamps and broken scrubbing-- Fixed tag input fields not selecting dropdown values on job application forms-- Fixed per-step video sync spawning redundant ffmpeg processes-- Fixed workflow parameter buttons hidden behind long parameter names-- Fixed quickstart failing on Windows due to missing `DATABASE_STRING`-- Fixed 1Password upstream failures showing unclear failure reasonsdiff --git a/docs/changelog-emails/2026-06-june.md b/docs/changelog-emails/2026-06-june.mddeleted file mode 100644index 1188e978a..000000000--- a/docs/changelog-emails/2026-06-june.md+++ /dev/null@@ -1,114 +0,0 @@---# Skyvern Changelog — June 2026--*Everything we shipped in June 2026 — weeks of June 1, 8, 15, and 22.*-------## Workflows Are Now Agents--We renamed **Workflows to Agents** across the product, the public API, and the docs. The API now lives under **`/agents`** (e.g. `POST /v1/run/agents`) and accepts an **`agent_id`** alias for `workflow_id` — same `wpid_` value, echoed back in responses. It's fully backwards-compatible: your existing `workflow_id` inputs, `/workflows` endpoints, SDK methods, and bookmarked doc URLs all keep working.---------## Workflow Studio (Preview)--Meet the next-generation editor. **Workflow Studio** is an opt-in preview that reimagines how you build: a refreshed **build canvas**, an **integrated live browser view**, and a built-in **run viewer** — all in one place. Take control of the live browser, center the viewport, and watch a run unfold without leaving the editor. Flip it on from Feature Preview in Settings.---------## Code-First Code Blocks--For when you want to drop to code. Code blocks now have a **dedicated code editor** with **syntax highlighting**, **Jinja parameter hints**, and a **run timeline** that records every action — including `page.evaluate(...)` calls. A **Build ↔ Code toggle** flips between visual and code editing, block labels and extraction output render correctly, and Copilot-generated code blocks support a **typed extraction schema** and fill logins with your stored credentials.---------## Copilot: Ask, Build, Code — With Memory--The Copilot got sharper about intent and context. It now **routes every request by mode** — **Ask** answers without touching your workflow, **Build** creates or edits it, **Code** generates executable code blocks. It **remembers the full chat history per workflow**, so prior context and decisions carry forward as you keep building. And its **repair loop is honest now**: when consecutive fixes make no verified progress, it stops after three tries and escalates cleanly — draft preserved — instead of spinning.---------## A Bigger Model Menu--More choice for every run. New this month: **DeepSeek**, **Xiaomi MiMo** (and **MiMo-V2.5** via OpenRouter), **Gemini 3.5 Flash**, **GPT-5-mini** (your own OpenAI key or via OpenRouter), and **Claude Opus 4.8** (Anthropic + Bedrock). OpenRouter runs now track token counts and cost like every other provider. (Gemini 2.5 Pro/Flash are deprecated; CUA and the Claude Opus family are now enterprise-only.)---------## Credentials & OTP, Everywhere--Logins that just work. Pull credentials straight from **1Password** (with an item picker) or **Bitwarden**, look up one-time passcodes from **Gmail** via OAuth, and **fill OTP at runtime** from your stored TOTP secrets — even inside code blocks. Expired credential sessions **auto re-save**, and you can **clear a saved credential** without deleting and recreating it.---------## Multi-Tab Browser Control--Automations that span tabs. The autonomous agent loop can now **open and work across multiple browser tabs** at once — following new-tab links and handling pop-ups — so flows that jump between tabs no longer dead-end.---------## Self-Host on Google Cloud--Full GCP-native deployments. Self-hosted Skyvern now supports **Google Cloud Storage** for artifacts and **GCP Secret Manager** as a vault backend — so you can run the whole stack on Google Cloud with your own storage and secrets.---------## Quick List — June--**New features**-- **Analytics drill-down & tag filters** — period-aware per-agent drill-down (avg duration, credits/run, period-over-period trends) plus filtering runs and metrics by workflow tag-- **Agent Directory Tree** — the agents/folders list is now a navigable tree for nested hierarchies-- **Self-serve editor onboarding** — a get-started modal, a 4-stop guided tour, and smart empty states for faster time-to-first-automation-- **Bulk actions on the agents list** — select multiple agents and act on them at once-- **Label management in Settings** — rename, recolor, and soft-delete tags-- **Browser profiles in MCP sessions** — associate a browser profile with MCP-driven runs for credential/state reuse-- **Feature Preview panel in Settings** — opt into upcoming features before they're fully released-- **Workflow schedules in open source** — self-hosted deployments now support schedules-- **Uncapped workflow runtime** — configure long-running automations with no maximum time limit--**Improvements**-- Structured output is verified against its schema before a run is marked complete-- Tasks return a best-effort **partial deliverable** when they exhaust their credit budget-- The task planner prefers **loop-based tasks** for repeated same-shaped work, and can now emit EXTRACT_INFORMATION / GOTO_URL actions-- The agent **acts immediately on stalled pages** and **halts reliably on terminal blockers** instead of spinning-- Remote browser sessions capped at 4 hours; debug sessions stay open after a run so you can inspect the final page-- End-state screenshots captured for **every open tab** at completion; per-run recordings clipped to the run window and stored as downloadable artifacts-- Run app + recording URLs surfaced in MCP and CLI responses; run-ID search fallback on the runs list-- Batched cache invalidation on workflow save (removed a per-block N+1) for faster saves on large workflows-- OpenRouter cost tracking; analytics tag filters accept labels, groups, or `group:label`-- Improved authenticator-2FA credential setup; PDF field mapping by positional labels-- Copilot data-write blocks default to `continue_on_failure=false`; code-repair progress shown as quiet indicators--**Bug fixes**-- Fixed the workflow builder freezing on an infinite FlowRenderer layout loop-- Fixed the code editor crashing (stack overflow) on large or deeply-nested JSON-- Fixed the app failing to load after a deploy from stale chunks — it now auto-recovers with a cache-busting reload-- Fixed login credentials not reattaching after editing a workflow via MCP-- Fixed conditional routing not being preserved when loading a v1-format workflow-- Fixed the active run context being lost when switching organizations-- Fixed multi-step logins not advancing when credentials filled JavaScript-gated fields-- Fixed scanned PDFs being parsed as a single page instead of page-by-page-- Fixed shadow-DOM text nodes being missed during element text extraction-- Fixed runs not stopping when a configured time limit was exceeded-- Fixed credits not recorded when data was captured inside a Navigate block-- Fixed telephone inputs dropping digits on bare NANP-format numbers-- Fixed persistent sessions reporting a stale cached session as live-- Fixed FileUpload blocks silently skipping when no file was provided (now a clean no-op)-- Fixed the block library drawer overflowing its container in the workflow builderdiff --git a/docs/changelog-emails/2026-07-july.md b/docs/changelog-emails/2026-07-july.mddeleted file mode 100644index 6294c1e81..000000000--- a/docs/changelog-emails/2026-07-july.md+++ /dev/null@@ -1,149 +0,0 @@---# Skyvern Changelog — July 2026--*Everything we shipped in July 2026 — weeks of June 30, July 6, 13, 20, and 27.*-------## Workflow Studio, for Everyone--The next-generation editor is out of preview — **Workflow Studio is now available to all users**, with a month of upgrades to match:--- **Build**: a redesigned layout with resizable, drag-to-reorder panes, plus a **full-screen YAML editing mode**-- **Runs**: a **Past Runs tab**, run **inputs and outputs broken down field by field**, code-block outputs shown inline, and searchable JSON-- **Polish**: block search, jump-to-start/end buttons, an unsaved-changes indicator that summarizes what changed, recent-activity history on the canvas, shareable block selection in the URL, and light mode---------## Human-in-the-Loop Approvals--When a step needs a human, you're in the loop. If a workflow step requires human review, you can now **approve or reject it directly from either run view** — no side channel required. The public run status API also reports a **`paused` state**, so your integrations get an accurate picture when a run pauses, including while it waits on review.---------## Code-First Browser Recording (Preview)--Show it once, get code. **Record actions in a live browser session** and Skyvern generates **reusable workflow code** from what you did — available now as an opt-in preview. Recording itself was redesigned around a **live draft panel** instead of a separate editor, with cleaner screenshots and more step types. **Live-draft enrichment is noticeably faster** too, so suggestions show up sooner while you record.---------## Credentials & Secrets, Leveled Up--Logins got a serious upgrade this month. The Credentials page adds a **Magic Link tab**, **passkey 2FA** for password credentials on enterprise plans, and an email-based 2FA identifier — point a credential at a connected inbox and Skyvern retrieves the code automatically.--Workflows got smarter about using them, too. They can **rotate through a pool of credentials** across runs and automatically retry with a fallback credential when one fails. Runs that share a credential are serialized, so they no longer clobber each other's login sessions. The builder's credential picker is searchable with no 100-item cap. And to keep secrets out of sight, a per-workflow **mask-secrets setting** hides sensitive values in run outputs and logs, while the HTTP Request block can fetch response values as secrets and pass them between blocks.---------## Document & File Pipelines--Files in, files out — anywhere. A new **Email Inbox block** (Gmail and Outlook) lets workflows read and act on inbox messages. A **Split PDF block** divides documents into separate files, and the File Parser now unpacks ZIP archives.--On the way out, the Download File block can deliver to **SFTP, Amazon S3, Azure, and Google Drive**, and SFTP joins the Cloud Storage upload targets. Google Drive uploads handle **files larger than 5 MB** with resumable transfers that pick up where they left off. The Cloud Storage block takes an optional prompt to upload only the files you want, and downloaded files get the correct extension detected from their content.---------## Run Tags & a Weekly Analytics Digest--Organize runs your way — and get the story delivered. Runs can now be **tagged**, manually or via **auto-detected platform tags**, managed in bulk, and filtered by agent or run type in Run History. Analytics gained **grouping by run metadata** and a restructured console for a clearer read on usage and cost. And a new **weekly email digest** lands in your inbox with run volume, credit usage, a status breakdown, week-over-week trend indicators, and your lowest-success-rate workflows.---------## Custom LLMs & New Models--Bring your own brain — or pick a new one. Organizations can now set **default Smart and Fast custom LLMs** from account settings instead of configuring them on every workflow. Custom LLM setup accepts **provider-specific parameters**, and Google Gemini is natively supported as a provider. The model menu grew too: **Claude Opus 5**, **Grok 4.5**, Gemini 3.5 Flash Lite, Gemini 3.6 Flash, and DeepSeek V4 Flash are all selectable now.---------## Self-Healing You Can See--Reliability, made visible. The workflows list now shows **reliability badges** when automatic recovery has kept a workflow on track. Run pages surface a **self-heal panel** showing exactly when recovery kicked in. And **self-healing can be switched on or off** right in workflow settings.--Under the hood, runs got harder to derail:--- Every browser action is bounded by a timeout, so a stuck action can't hang a run-- Workflows **wait through virtual queues and waiting-room pages** instead of failing-- Cloudflare Turnstile challenges are detected proactively for faster resolution, and simple arithmetic text CAPTCHAs hit during data lookups are solved automatically-- Code-block self-healing enforces a minimum retry floor before giving up---------## Quick List — July--**New features**--- **In-app MCP setup** — configure the MCP connector directly inside the app, plus new MCP endpoints for **1Password and Bitwarden**-- **Google integrations with org-level OAuth** — centralized admin control over connected Google accounts-- **Browser session startup URL** — browser sessions can start at a specific URL-- **Recurring add-on credits** — set up monthly add-on credits on your subscription, with the amount configurable from billing-- **Persistent credential vault for self-hosted Docker** — saved credentials now survive restarts--**Improvements**--- MCP sessions gained a **live action timeline** and an artifact timeline tab-- MCP tool results now lead with the app + recording URLs so you can jump straight in-- MCP observe surfaces custom dropdown options that were previously hidden-- Large workflows are easier to read and safely edit through MCP-- Copilot follows test runs **live in the Live Browser**, narrates studio runs as they execute, and reveals its verification actions progressively-- Copilot **pauses to ask for a credential** when one is needed and surfaces a credential card right in chat-- Copilot auto-matches saved credentials to the login site and can read a one-time sign-in code from an email inbox-- Redesigned Copilot composer with a clearer mode indicator-- Copilot handles login and verification pages more safely mid-build, and says so honestly when a turn runs out of time-- Buy additional credits on demand from the billing page — and recurring add-on credits are cheaper at **$0.90 per 1,000**-- OTP parsing usage is now visible on the billing page-- Persistent browser sessions renew their lease automatically with activity-- Save & Reuse Session runs use a **consistent egress IP** and are backed by managed browser profiles-- Browser profiles can be refreshed in place, show role badges, and are protected from deletion while in use-- Loop block iteration limit raised from **500 to 1,000**-- The Cloud Storage block now handles up to **300 files** per run-- Google Sheets spreadsheet and sheet pickers accept **template expressions** for dynamic destinations-- Select-option now works on **custom and autocomplete-style dropdowns**, not just native selects-- Scheduling starts with a **guided date/time picker**-- The editor highlights run-blocking blocks and disables Run until they're resolved-- Improved extraction accuracy when requested data is **genuinely absent** from the page-- Failed Code blocks now include a **screenshot and the final page URL**, and failure messages surface the actual exception and failing line-- Run history loads much faster for workflows with many runs-- Expired integrations re-authenticate **in place**, without reconnecting from scratch-- Run views make screenshots easier to find, show output values more clearly, and display conditional/loop prompts in block settings-- Read-only workflow canvases let you click through conditional branches-- Login blocks prefill their goal from saved credential instructions and support an action history option-- The Labels page adds search and protects built-in system tags-- Files delivered via inline data URLs are captured as run files-- ZIP parsing always returns the list of extracted files--**Bug fixes**--- Fixed issues affecting automated solving of Cloudflare and reCAPTCHA challenges, including reCAPTCHA v2 being treated as solved before its token was ready-- The analytics dashboard now shows credit usage from actual billing records instead of an estimate, so the numbers match your invoice-- Fixed a race condition that could corrupt credential values — rotating or dropping characters — as they were typed into login fields-- Fixed a regression where downloaded files could lose their original filenames, and files being uploaded under a temporary name before the browser finished renaming them-- Fixed phone numbers being altered or rejected when country-aware form fields rewrote them with a country code-- Fixed Cmd+V paste not working in the Live Browser view on macOS-- Fixed screenshots, viewers, and artifact download links expiring — they now refresh automatically-- Fixed importing workflow exports bundled as multi-workflow YAML/JSON files-- Fixed the run detail page returning a 404 for runs based on a template workflow-- Fixed LLM blocks not enforcing their configured output schema, which could let malformed data reach downstream blocks-- Fixed custom segmented date inputs silently submitting the wrong month-- Fixed a valid "no" answer from a workflow causing an otherwise successful run to be marked as failed-- Fixed the Live Browser view showing blank frames during a run-- Fixed missed download popups causing file-download clicks to silently fail-- Fixed completed Copilot runs being incorrectly recorded as failures-- Fixed failures inside a Finally block not being propagated to the run outcome-- Fixed a Python comment causing an entire Code block to faildiff --git a/docs/changelog-emails/2026-08-august.md b/docs/changelog-emails/2026-08-august.mddeleted file mode 100644index b80b78102..000000000--- a/docs/changelog-emails/2026-08-august.md+++ /dev/null@@ -1,183 +0,0 @@---# Skyvern Changelog — August 2026--*Everything we shipped in August 2026 — weeks of August 3, 10, 17, and 24.*-------## Workflow Copilot Updates--The updated Copilot chat is now the default everywhere, including the workflow editor and Workflow Studio.--- Build progress uses a **flat activity log** of what Copilot looked at, wrote, and ran. Write steps show code diffs. Narrator lines explain steps in place.-- Copilot can **repair and rerun one block** in a single step. If the browser remains on that block's page, Copilot reruns only that block without replaying the login. After a code block fails, Copilot receives the recorded browser error. It also receives the screenshot and any blocking overlay when they were captured.-- The Accept card lists added, changed, and unchanged blocks. Accept and Reject appear without a reload. The new **Test end-to-end** action runs the whole draft in a fresh browser.-- Select a connected Google account inline or choose credentials from a **searchable picker**. You can give Copilot a saved login by name. Copilot solves captchas and reads sign-in codes from a connected inbox without stopping to ask.--Copilot supports the human-approval block and knows today's date. It retains gathered page facts across long sessions. The new stop button ends the current Copilot run.---------## Skyvern 3.0 Engine (Beta)--The opt-in **Skyvern 3.0** beta controls the browser through a persistent tool loop. Only actions count against your step budget. Perception turns do not count.--- Select Skyvern 3.0 (marked Beta) in the engine selector for task, navigation, login, action, and extraction blocks. You can also pass `engine="skyvern-3.0"` on a task run.-- When the page's text digest has gaps, the agent takes a **marked-up screenshot** and acts on its numbered marks.-- The engine solves captchas, including hCaptcha, and resolves **TOTP and emailed one-time codes**. It follows emailed magic links without those links reaching the model.-- Custom comboboxes, click-to-open selects, typeaheads, and shadow-DOM controls commit in **one action**. The engine verifies file uploads and toggle changes. It waits for the page to settle before completing a block.--The engine supports run replay and typed action history.---------## Control Chrome through MCP--The new **Skyvern Agent** Chrome extension is available on the Chrome Web Store. It lets Skyvern MCP control your logged-in Chrome through `BROWSER_TYPE=extension-connect`. This includes existing sessions, cookies, and profiles. You share only the tabs you choose. Pairing requires one approval and persists per workstation. Several coding agents can share one Chrome at the same time.--You can select MCP scopes at install time or on the hosted URL.--- Set `--scope all|operate|build|browser` during installation. You can also opt into the **`lean`** scope for browser-only tools.-- The `lean` and `all` scopes include new tools. `skyvern_page` reads large pages with cursor pagination. `skyvern_extract_structured` returns schema-validated results. `skyvern_finish` declares a task's final outcome.-- Add Skyvern as a custom connector in **Claude.ai and Claude Desktop** through hosted MCP. Refresh tokens renew access without prompting you. Hosted MCP connects to the organization you picked at sign-in.---------## Credentials and Login Updates--The Login block now lists 1Password items in its credential dropdown.--- Bind a **1Password** item directly from the Login block's credential dropdown. A five-step guide explains the service account token. Items that store one-time codes in a plain field can specify a custom TOTP field.-- **Passkey 2FA** is available for every cloud organization. You can use Outlook as a source for email two-factor codes.-- Store a credential with a username and no password. Leave a workflow credential's default blank to **run unauthenticated unless you pass one at run time**.-- Run workflows **sequentially per credential**, including scheduled runs. Save & Reuse Session keeps one browser profile per rotating credential.--Skyvern stores magic-link and OTP values exactly as received. Code blocks and cached scripts can complete an emailed sign-in with the new `magic_link` verb. Bitwarden updates reduce credential fetch time and failures. OTP logins use their full 15-minute polling budget.---------## Files, Email & Storage--S3-compatible storage endpoints must use public HTTPS by default. Self-hosted deployments can allow internal hosts.--- Set the new **`endpoint_url`** field on File Upload and File Download blocks to use your S3-compatible storage.-- Configure **Custom SMTP** under Advanced Settings to send from your own mail server through the Send Email block.-- Set **`retention_days`** when uploading a file with `POST /v1/upload_file`. Delete any upload with `DELETE /v1/files/{id}`. Attach uploads to a run with `file_ids` to have Skyvern delete them when the run ends.-- File Download uses your connected Drive account for private Drive files. The Drive folder ID is optional. Leave it blank to save to My Drive root. Google Sheets appends use the column you specify. Sheets, Drive, Docs, and Gmail blocks resolve connections **by name or email**.--Inside a loop, the File Parser can skip files it cannot parse. It follows Drive links behind confirm pages. A slow parse of a large document fails only its own block, not the whole run.---------## Browser Sessions & Live View--Workflow runs can share one browser while retaining logged-in state and tabs.--- Enable the new **`reuse_browser_session`** workflow setting. Each run can override the setting.-- Externally hosted browsers (US cloud) and session-backed runs now stream a **live view**. Take-control reaches the full page height. The view reports when its stream stops instead of freezing.-- The API **caps over-limit timeouts at 240 minutes** and reports them in a `warning` field instead of rejecting them. A closed session reports its status and what to do next.--Recordings from external providers are retained after teardown. A renewal never shortens a session.---------## Code Block Errors and Outputs--Run status and webhooks identify typed business errors from code blocks.--- Declare **`error_code_mapping`** on a code block. Use `raise ErrorCode("out_of_stock", "why")` to identify the failed condition.-- A failed block reports the failing line and a plain-language next step. It includes the raised exception when redaction allows. User-code and browser failures link to the **block screenshot** when one was captured.-- Run detail displays **persisted code-block output** instead of a disabled Outputs panel.-- The sandboxed runner supports downloads, `page.url`, `Locator.or_`/`and_`, Python `class` statements, `page.frames`, and your configured timeouts. Advisory **security-scan warnings** appear when you save code in Copilot or lint through MCP.---------## Analytics & Run History--The run failure rate now excludes runs you cancelled.--- **Group by workflow tag** on the analytics dashboard for per-tag run and credit chargeback. The existing run-metadata grouping remains available.-- An **Actions** tile shows action counts and failure rate on the dashboard and workflow health panel. The run chart switches to a **Failures view** with agents ranked by failures first.-- The selected **run status** applies to every chart, tile, table, and CSV export, including drill-downs. You can select several statuses at once.-- Run history search and status filters persist in the URL. `GET /runs` accepts the new `failure_category` filter for workflow runs.---------## Quick List — August--**New features**--- `GET /v1/workflows/{workflow_id}/runs` and the MCP workflow run list accept `include_child_runs` to list child runs.-- Send Email works through the API, SDK, and MCP without declaring platform SMTP parameters for `send_email` blocks.-- Export several agents as a ZIP with one file per agent.-- Open a workflow's schema/YAML from the header menu in the classic workflow editor.-- Self-hosted deployments without a cron worker can enable an in-process reaper for stuck runs.-- Self-hosted deployments now solve Cloudflare challenges through the bundled solver extension.--**Improvements**--- A card shows Workflow Inputs where the flow starts. Its header shows the input count.-- Studio makes pane borders more distinct and simplifies timeline controls. The Run Timeline shows elapsed time, blocks, actions, and credits.-- Paste a full workflow export into the editor's YAML pane.-- New code blocks from browser recording, Browser Task steps, and PDF/SOP imports open in the new code block editor.-- Delete confirmations use the same dialog across the app.-- Run detail pages no longer freeze on multi-megabyte output values.-- A workflow run with no blocks fails immediately with a specific reason. An undefined for-loop parameter returns a 422.-- Ordinary text inputs fill atomically. Toggles and checkboxes use their intended state.-- Dropdown and custom-select fills verify that the value committed instead of reporting a false success.-- Runs recover from CDN access-denied interstitials and can click data-grid row checkboxes.-- When a large page causes prompt overflow, the step trims the element tree instead of failing.-- Proxied stealth browser launches start about 30 seconds faster. Timezone and locale default to the proxy's region unless you set them.-- Job-application fills on a major applicant-tracking platform no longer fabricate values. They commit dates and pickers correctly and recognize submit confirmation. They stop early when a posting shows you already applied.-- Revoked or rotated API keys stop working within 60 seconds.-- The CLI refuses to send an API key to the cloud default unless `SKYVERN_BASE_URL` is set explicitly.-- The API caps concurrent SDK `run_action` calls per organization with a 429 to prevent degraded service for everyone.-- The API returns 503 with `Retry-After` when the database connection drops, so SDK clients retry.-- MCP tools return a structured `SESSION_EXPIRED` result. Large MCP responses are compacted while keeping identifiers and errors.-- The standalone MCP server binds to localhost by default (`--host 0.0.0.0` to expose it).-- Connecting a Google integration requests only that integration's scopes.--**Bug fixes**--- Fixed scheduled runs failing for workflows whose credential is set to run sequentially.-- Fixed a workflow parameter that matched a credential being typed into the page as `[REDACTED_SECRET]`.-- Fixed long text typed into free-text fields keeping only its last ten characters.-- Fixed saved browser profiles that failed to load, dropped the sign-in, or showed the "Restore pages?" bubble.-- Fixed magic-link and OTP values changing before use. Fixed Gmail sign-in link emails failing to parse.-- Fixed Bitwarden logins without a two-factor secret failing to list or fill.-- Fixed downloads lost on adopted, vendor-hosted, and reused browser sessions. Fixed lost downloads from in-page PDF blobs and after a download closed the working page.-- Fixed a downloaded file being materialized twice and uploaded as duplicates.-- Fixed an aborted browser download reporting success with zero files.-- Fixed webhook deliveries leaving from a different IP than the documented relay.-- Fixed SDK and MCP actions against cloud browser sessions returning HTTP 500.-- Fixed live view flickering to black and showing a black screen on session-backed runs. Fixed frames stopping after a reconnect.-- Fixed Delete and Backspace not working in the hosted browser live view.-- Fixed browser session recordings failing to save after long sessions.-- Fixed runs stuck in `queued` after a sequential gate deadline. Fixed runs retrying indefinitely when the workflow body raised an exception.-- Fixed saving a workflow with many cached scripts hanging for minutes.-- Fixed selecting a block marking the workflow as unsaved. Fixed the Inputs card opening Workflow Settings.-- Fixed "Execute on any outcome" blocks being marked as disconnected or skipped by a conditional branch.-- Fixed run and task view tabs blanking the content area. Fixed shared Studio run links opening the wrong run.-- Fixed the Add Credential modal not opening from the Credentials page.-- Fixed the analytics dashboard counting cancelled runs as failures. Fixed enterprise plans seeing always-zero credit tiles.-- Fixed self-serve customers being able to start a duplicate subscription checkout.-- Fixed self-hosted deployments failing to launch browsers, initialize encryption, or authenticate the UI behind a proxy.-- Fixed the `skyvern doctor` check not recognizing `ENABLE_OPENAI_COMPATIBLE`.-- Patched third-party dependency vulnerabilities.diff --git a/docs/changelog-emails/images/2026-04-01-copilot-v2.png b/docs/changelog-emails/images/2026-04-01-copilot-v2.pngdeleted file mode 100644index 074d3f03f..000000000Binary files a/docs/changelog-emails/images/2026-04-01-copilot-v2.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-04-02-mcp-oauth.png b/docs/changelog-emails/images/2026-04-02-mcp-oauth.pngdeleted file mode 100644index 1c4973d6e..000000000Binary files a/docs/changelog-emails/images/2026-04-02-mcp-oauth.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-04-03-cli-schedule-config.png b/docs/changelog-emails/images/2026-04-03-cli-schedule-config.pngdeleted file mode 100644index b54762995..000000000Binary files a/docs/changelog-emails/images/2026-04-03-cli-schedule-config.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-04-04-ai-summarize.png b/docs/changelog-emails/images/2026-04-04-ai-summarize.pngdeleted file mode 100644index bdb09c104..000000000Binary files a/docs/changelog-emails/images/2026-04-04-ai-summarize.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-04-05-error-code-mapping.png b/docs/changelog-emails/images/2026-04-05-error-code-mapping.pngdeleted file mode 100644index 18e8f16d4..000000000Binary files a/docs/changelog-emails/images/2026-04-05-error-code-mapping.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-04-06-debugger-run-dialog.png b/docs/changelog-emails/images/2026-04-06-debugger-run-dialog.pngdeleted file mode 100644index 292ba3174..000000000Binary files a/docs/changelog-emails/images/2026-04-06-debugger-run-dialog.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-05-00-header.png b/docs/changelog-emails/images/2026-05-00-header.pngdeleted file mode 100644index ad4b50be4..000000000Binary files a/docs/changelog-emails/images/2026-05-00-header.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-05-01-draggable-editor.png b/docs/changelog-emails/images/2026-05-01-draggable-editor.pngdeleted file mode 100644index 1312fcd12..000000000Binary files a/docs/changelog-emails/images/2026-05-01-draggable-editor.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-05-02-run-page-redesign.png b/docs/changelog-emails/images/2026-05-02-run-page-redesign.pngdeleted file mode 100644index ad2c8a746..000000000Binary files a/docs/changelog-emails/images/2026-05-02-run-page-redesign.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-05-03-while-loop.png b/docs/changelog-emails/images/2026-05-03-while-loop.pngdeleted file mode 100644index b0193660d..000000000Binary files a/docs/changelog-emails/images/2026-05-03-while-loop.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-05-04-analytics-dashboard.png b/docs/changelog-emails/images/2026-05-04-analytics-dashboard.pngdeleted file mode 100644index 2684c2c24..000000000Binary files a/docs/changelog-emails/images/2026-05-04-analytics-dashboard.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-05-05-copilot.png b/docs/changelog-emails/images/2026-05-05-copilot.pngdeleted file mode 100644index 185c4a654..000000000Binary files a/docs/changelog-emails/images/2026-05-05-copilot.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-05-06-run-control.png b/docs/changelog-emails/images/2026-05-06-run-control.pngdeleted file mode 100644index c4f6ec5a1..000000000Binary files a/docs/changelog-emails/images/2026-05-06-run-control.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-05-07-self-hosting.png b/docs/changelog-emails/images/2026-05-07-self-hosting.pngdeleted file mode 100644index 124c79c24..000000000Binary files a/docs/changelog-emails/images/2026-05-07-self-hosting.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-05-08-browser-profiles.png b/docs/changelog-emails/images/2026-05-08-browser-profiles.pngdeleted file mode 100644index a95f2ec8c..000000000Binary files a/docs/changelog-emails/images/2026-05-08-browser-profiles.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-06-00-header.png b/docs/changelog-emails/images/2026-06-00-header.pngdeleted file mode 100644index 64edf83c0..000000000Binary files a/docs/changelog-emails/images/2026-06-00-header.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-06-01-agents-rename.png b/docs/changelog-emails/images/2026-06-01-agents-rename.pngdeleted file mode 100644index b59997963..000000000Binary files a/docs/changelog-emails/images/2026-06-01-agents-rename.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-06-02-workflow-studio.png b/docs/changelog-emails/images/2026-06-02-workflow-studio.pngdeleted file mode 100644index c5bf24849..000000000Binary files a/docs/changelog-emails/images/2026-06-02-workflow-studio.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-06-03-code-first.png b/docs/changelog-emails/images/2026-06-03-code-first.pngdeleted file mode 100644index 9bf2eac49..000000000Binary files a/docs/changelog-emails/images/2026-06-03-code-first.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-06-04-copilot.png b/docs/changelog-emails/images/2026-06-04-copilot.pngdeleted file mode 100644index 4d6c1298a..000000000Binary files a/docs/changelog-emails/images/2026-06-04-copilot.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-06-05-models.png b/docs/changelog-emails/images/2026-06-05-models.pngdeleted file mode 100644index 8266e172b..000000000Binary files a/docs/changelog-emails/images/2026-06-05-models.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-06-06-credentials.png b/docs/changelog-emails/images/2026-06-06-credentials.pngdeleted file mode 100644index a50398a5d..000000000Binary files a/docs/changelog-emails/images/2026-06-06-credentials.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-06-07-multi-tab.png b/docs/changelog-emails/images/2026-06-07-multi-tab.pngdeleted file mode 100644index 038d99f6e..000000000Binary files a/docs/changelog-emails/images/2026-06-07-multi-tab.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-06-08-gcp-self-host.png b/docs/changelog-emails/images/2026-06-08-gcp-self-host.pngdeleted file mode 100644index 15338f88b..000000000Binary files a/docs/changelog-emails/images/2026-06-08-gcp-self-host.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-07-00-header.png b/docs/changelog-emails/images/2026-07-00-header.pngdeleted file mode 100644index 66bf5bf30..000000000Binary files a/docs/changelog-emails/images/2026-07-00-header.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-07-01-workflow-studio-ga.png b/docs/changelog-emails/images/2026-07-01-workflow-studio-ga.pngdeleted file mode 100644index 3799041c7..000000000Binary files a/docs/changelog-emails/images/2026-07-01-workflow-studio-ga.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-07-02-hitl-approvals.png b/docs/changelog-emails/images/2026-07-02-hitl-approvals.pngdeleted file mode 100644index c58ff042f..000000000Binary files a/docs/changelog-emails/images/2026-07-02-hitl-approvals.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-07-03-browser-recording-code.png b/docs/changelog-emails/images/2026-07-03-browser-recording-code.pngdeleted file mode 100644index 5cf0c87df..000000000Binary files a/docs/changelog-emails/images/2026-07-03-browser-recording-code.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-07-04-credentials.png b/docs/changelog-emails/images/2026-07-04-credentials.pngdeleted file mode 100644index 40e3af16e..000000000Binary files a/docs/changelog-emails/images/2026-07-04-credentials.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-07-05-file-pipelines.png b/docs/changelog-emails/images/2026-07-05-file-pipelines.pngdeleted file mode 100644index a2584aac2..000000000Binary files a/docs/changelog-emails/images/2026-07-05-file-pipelines.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-07-06-run-tags-digest.png b/docs/changelog-emails/images/2026-07-06-run-tags-digest.pngdeleted file mode 100644index 1c91de474..000000000Binary files a/docs/changelog-emails/images/2026-07-06-run-tags-digest.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-07-07-custom-llms-models.png b/docs/changelog-emails/images/2026-07-07-custom-llms-models.pngdeleted file mode 100644index de3da2c91..000000000Binary files a/docs/changelog-emails/images/2026-07-07-custom-llms-models.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-07-08-self-healing.png b/docs/changelog-emails/images/2026-07-08-self-healing.pngdeleted file mode 100644index 45e6a8857..000000000Binary files a/docs/changelog-emails/images/2026-07-08-self-healing.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-08-00-header.png b/docs/changelog-emails/images/2026-08-00-header.pngdeleted file mode 100644index cf972317b..000000000Binary files a/docs/changelog-emails/images/2026-08-00-header.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-08-01-copilot-updates.png b/docs/changelog-emails/images/2026-08-01-copilot-updates.pngdeleted file mode 100644index e587fe9f5..000000000Binary files a/docs/changelog-emails/images/2026-08-01-copilot-updates.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-08-02-skyvern-3-engine.png b/docs/changelog-emails/images/2026-08-02-skyvern-3-engine.pngdeleted file mode 100644index f72780135..000000000Binary files a/docs/changelog-emails/images/2026-08-02-skyvern-3-engine.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-08-03-control-chrome-mcp.png b/docs/changelog-emails/images/2026-08-03-control-chrome-mcp.pngdeleted file mode 100644index 7b61e7d13..000000000Binary files a/docs/changelog-emails/images/2026-08-03-control-chrome-mcp.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-08-04-credentials-login-updates.png b/docs/changelog-emails/images/2026-08-04-credentials-login-updates.pngdeleted file mode 100644index 602f96e77..000000000Binary files a/docs/changelog-emails/images/2026-08-04-credentials-login-updates.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-08-05-files-email-storage-blocks.png b/docs/changelog-emails/images/2026-08-05-files-email-storage-blocks.pngdeleted file mode 100644index 3905ac4b2..000000000Binary files a/docs/changelog-emails/images/2026-08-05-files-email-storage-blocks.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-08-06-browser-sessions.png b/docs/changelog-emails/images/2026-08-06-browser-sessions.pngdeleted file mode 100644index b55102001..000000000Binary files a/docs/changelog-emails/images/2026-08-06-browser-sessions.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-08-07-code-block-errors-outputs.png b/docs/changelog-emails/images/2026-08-07-code-block-errors-outputs.pngdeleted file mode 100644index 1a8c429f6..000000000Binary files a/docs/changelog-emails/images/2026-08-07-code-block-errors-outputs.png and /dev/null differdiff --git a/docs/changelog-emails/images/2026-08-08-analytics-run-history-filters.png b/docs/changelog-emails/images/2026-08-08-analytics-run-history-filters.pngdeleted file mode 100644index 352eb6412..000000000Binary files a/docs/changelog-emails/images/2026-08-08-analytics-run-history-filters.png and /dev/null differdiff --git a/docs/changelog-emails/templates/changelog-header-template.png b/docs/changelog-emails/templates/changelog-header-template.pngdeleted file mode 100644index 60018e2e7..000000000Binary files a/docs/changelog-emails/templates/changelog-header-template.png and /dev/null differdiff --git a/docs/design/2026-06-15-workflow-build-recent-activity-design.md b/docs/design/2026-06-15-workflow-build-recent-activity-design.mddeleted file mode 100644index e94bbfc5d..000000000--- a/docs/design/2026-06-15-workflow-build-recent-activity-design.md+++ /dev/null@@ -1,306 +0,0 @@-# Workflow Build — Recent Activity History (SKY-11075)--Replace the bottom-of-canvas **dot/carousel indicator** on `/workflows/:id/build` with a-real recent-activity history view that's easy to scan and lets you jump directly to a-specific recent activity.--Status: **Final — recent activity is a compact run selector locked to the right-run-detail panel; the A/B/C style and placement preview toggles were removed. See-the Round 7 section below.**-------## 1. Frame--- **Page:** Workflow editor, *build* mode — `/workflows/:id/build` (and the run-loaded- variant `/workflows/:wpid/:workflowRunId/:blockLabel/build`). The history surface only- renders in **debug/browser mode** (`showBrowser`), bottom-center over the canvas+code- pane. Redesign of an existing affordance.-- **Job to be done:** While iterating on a workflow you fire blocks/the workflow many times- in one debug session. Today those runs collapse into a row of identical ~16px dots whose- only metadata is a hover tooltip. You can't tell at a glance *what* ran, *whether it- passed*, *how long it took*, or *which mode* (agent vs cached code) — you must hover each- dot one by one. The job: make each recent run legible at a glance and jump straight to it.-- **Deciding constraint — what counts as "recent activity":** **Debug-session runs**- (confirmed with requester). Each entry = one block/workflow run triggered in the current- debug session, sourced from `GET /debug-session/{id}/runs` (`useDebugSessionRunsQuery`).- No new backend. Deliberately distinct from (a) the global **Runs** page and (b) the- right-side **run-detail timeline** (`DebuggerRun`/`DebuggerRunTimeline`) which shows the- *inside* of one selected run. This surface is the *list of runs*; selecting one drives the- existing detail rail — no overlapping history surfaces.-- **Open sub-question (requester):** is a dock/strip the best surface, or do we want a- "proper timeline"? → resolved via the concepts below; requester to pick direction.-- **Platform:** web.-- **North star references:** Databricks pipeline run-selector + event log; Vercel Activity feed.--### Data available per entry (`DebugSessionRun`) — today all hidden behind one dot-`block_label`, `status` (running/completed/failed/terminated/timed_out/canceled/queued/-skipped), `created_at` / `queued_at` / `started_at` / `finished_at` (→ duration + "x ago"),-`failure_reason`, `run_with` (agent | code), `code_gen`, `ai_fallback`, `workflow_run_id`,-`workflow_permanent_id`, `script_run_id`. Selecting → `navigate(/workflows/{wpid}/{workflow_run_id}/{block_label}/build)`.-------## 2. DS inventory (read from code — Figma DS is thin by design)--| Need | Reuse (exists in repo) |-|---|---|-| Status glyph | `StatusDot` pattern (WorkflowRunTimelineBlockItem): `CheckCircledIcon text-success`, `CrossCircledIcon text-destructive`, `ReloadIcon animate-spin text-sky-400`, neutral `rounded-full bg-slate-600` |-| Block-type glyph | `WorkflowBlockIcon` (per `WorkflowBlockType`) |-| Type / mode pills | inline pill convention `rounded bg-slate-700/70 px-1.5 py-0.5 text-[10px] font-medium` |-| Relative time / duration | `formatMs(Date.now()-created).ago`, `formatDuration(toDuration(ms))`, `tabular-nums` |-| Tooltip | `Tip` (`@/components/Tip`) |-| Scroll container | `ScrollArea` (`@/components/ui/scroll-area`) |-| Dropdown / flyout | `Popover` (`@/components/ui/popover`) |-| Expand / collapse | `Collapsible` (`@/components/ui/collapsible`) |-| Buttons / badges | `Button`, `Badge` |-| Surface tokens | `bg-neutral-50 dark:bg-background`, `border-neutral-200 dark:border-neutral-800`, focus `focus-visible:ring-1 focus-visible:ring-white/40` |--Conventions: build-mode rails skew dark (`slate-*`); pills `text-[10px]`, `tabular-nums` for-numbers; every interactive row is a `<button>` with a visible focus ring.--## 3. Gap ledger (DS lacks → candidate Step-5 tickets)-- No dedicated **Status badge** component — status color/icon mapping is re-derived inline in- several places (`StatusDot`, run cards). Candidate: extract a shared `RunStatusBadge`.-- No **horizontal "activity strip"/run-history rail** primitive.-- (Append as concepts/implementation pressure the library.)-------## 4. Inspiration (Mobbin — web)--1. **Databricks — pipeline "Update details"** — editor canvas with a **Run: {timestamp}- [status]** selector top-left, a detail panel (Update ID, *Status, Duration, Start time,- Run time, Completion date*), and a timestamped **Event log** below. Closest analog: a- build canvas whose run history is a compact selector + per-run metadata.- [link](https://mobbin.com/screens/2a53f9eb-b8e1-4958-afb7-f441acbdf414). a11y: the- selector must be a real listbox/menu button, not a click-only div.-2. **Vercel — Activity** — scannable vertical feed: each row = entity icon + plain-language- description + **relative time** ("42m"), grouped by date, left filter rail. Gold standard- for "legible at a glance."- [link](https://mobbin.com/screens/2f3056eb-a31a-4b3e-ae4b-1bcc1352117f). a11y: relative- time needs a `title`/`<time datetime>` with the absolute timestamp.-3. **WorkOS — Events** — master-detail: list (event + date/time) on the left, metadata on the- right. [link](https://mobbin.com/screens/a5b50327-60e0-4075-9cfd-7f383d18265c).-4. **Okta — System Log** / **Stripe — Events** / **Cloudflare — Observability** — dense log- tables (timestamp · type · message), some with a count-over-time bar.- [Okta](https://mobbin.com/screens/70f42b3f-1823-4cb2-872d-346ad3eab8dc) ·- [Stripe](https://mobbin.com/screens/5423e185-8374-4b6c-a4a0-2d3ae9adc1d0) ·- [Cloudflare](https://mobbin.com/screens/db64bfde-eac6-4786-8c29-62b98eb1a149). a11y:- color-coded status must pair with an icon/text, not color alone.-------## 5. Concepts--All three keep the data scope (debug-session runs), keep the existing bottom-center anchor-over the canvas+code pane, and preserve the existing select→navigate behavior. They differ in-*how the history is surfaced*.--### Concept A — "Activity Strip" (in-place horizontal chips)-- **Thesis:** smallest change. Swap the row of bare dots for a row of legible **chips** in the- same pill, scrolling horizontally. Each chip: status glyph + block icon + block label +- "x ago"; duration/mode in tooltip. Click → navigate.-- **Lineage:** Databricks run-selector (canvas-anchored) + Vercel chip legibility.-- **DS:** `StatusDot`, `WorkflowBlockIcon`, pills, `ScrollArea`, `formatMs`. Same wrapper/position.-- **Tradeoffs:** + tiny diff, preserves muscle memory, fast. − horizontal scroll is less- scannable than vertical; limited metadata per chip; long lists hide off-screen.--### Concept B — "Activity History panel" (collapsed timeline bar → expanding vertical list) — RECOMMENDED-- **Thesis:** the dock's **collapsed** state is a slim **status-segment bar** — one thin- segment per run, colored by status, newest on the right, labeled "N runs · last ✓ 2m ago".- That's the at-a-glance, time-ordered "timeline." Click/hover **expands upward** into a- **vertical, scannable list**: one rich row per run = status glyph · block icon · block name ·- type pill · mode pill (Agent/Code) · duration · "x ago", with a one-line failure reason on- failures. Newest first. Row click → navigate (existing behavior); current run highlighted.-- **Lineage:** Vercel Activity feed (legible rows) + Okta count-over-time bar (collapsed- status density) + WorkOS master-detail (list feeds the existing right-side detail rail).-- **DS:** `Popover`/`Collapsible` + `ScrollArea`, the `WorkflowRunTimelineBlockItem` row- vocabulary, `StatusDot`, pills. New: a small **status-segment bar** (gap-ledger item).-- **Tradeoffs:** + most legible (vertical rows, full metadata — directly satisfies the AC),- + collapsed footprint stays tiny so the canvas isn't crowded, + serves *both* the- "scannable history" and "proper timeline" instincts, + scales to long lists.- − one extra primitive (segment bar); − popover sizing over the canvas needs care.--### Concept C — "Run Selector + Segmented Timeline" (header selector + bottom bar)-- **Thesis:** Databricks-style. A **run selector** near the header ("Run: Login ✓ · 2m ago ▾")- opens the activity list, *and* a persistent thin **segmented timeline bar** sits at the- bottom (each run a proportional, status-colored segment; hover→tooltip, click→navigate).-- **Lineage:** Databricks run-selector + Okta status-density bar.-- **DS:** `Popover`, segment bar (new), `StatusDot` colors.-- **Tradeoffs:** + strongest "timeline" metaphor, scales well. − most net-new UI and two- surfaces (header + bottom) risk fragmenting the history; − largest build/QA; − header is- already busy in build mode.--## 6. Recommendation & Decision--**Recommend Concept B.** It's the most faithful to the acceptance criteria — each entry-legible at a glance with real metadata, and one click to jump to a run — while keeping the-collapsed footprint as small as today's dot pill. Its collapsed status-segment bar answers the-requester's "proper timeline?" instinct without splitting the history across two surfaces-(Concept C's main risk), and it reuses the established run-row vocabulary so it looks native.--Decision (2026-06-15): requester chose to **build all three (A, B, C) behind a live variant-toggle** so they can compare directly in the PR preview, then collapse to one winner later.-Implementation ships a `RecentActivity` orchestrator that renders the variant selected by a-persisted (localStorage + `?activityVariant=` URL seed) toggle; default = B.--## 7. Implementation (shipped)--New module `skyvern-frontend/src/routes/workflows/debugger/recentActivity/`:-- `useRecentActivity.ts` — data hook (debug-session runs, sorted asc; block-type map;- `navigateToRun` preserving the old select→navigate target, now also resolving nested loop- blocks for the "block still exists" guard).-- `runActivity.ts` / `RunGlyphs.tsx` — shared status/duration/time helpers + status/block glyphs.-- `RecentActivitySegmentBar.tsx`, `RecentActivityList.tsx` — shared primitives.-- `RecentActivityStrip.tsx` (A), `RecentActivityPanel.tsx` (B), `RecentActivityTimeline.tsx`- (C), `RecentActivityVariantToggle.tsx`, `RecentActivity.tsx` (orchestrator).-- `src/store/RecentActivityVariantStore.ts` — variant store.-- `Workspace.tsx` now renders `<RecentActivity />`; `DebuggerBlockRuns.tsx` deleted.--Verification: `tsc --noEmit` ✓, `eslint --max-warnings 0` ✓, `vite build` ✓. Live visual-verification of the populated state needs a debug session with ≥1 run (backend + browser-session) — delegated to the PR preview by design (the whole point of the A/B/C toggle).--### Gap ledger outcome (→ Step 5 DS ticket)-- Shipped one-off **`RecentActivitySegmentBar`** (status-segment/timeline-bar primitive) — DS has none.-- Re-derived **run-status → icon/color** inline (`RunGlyphs` + `STATUS_PILL_TONE`/`SEGMENT_FILL`)- because there's no shared `RunStatusBadge`.-- Filed **SKY-11082** (team Skyvern AI) to extract both primitives into the DS once the winning- variant is chosen. Pure helpers covered by `runActivity.test.ts`.-------# Round 2 — Placement exploration (2026-06-16)--The A/B/C work settled the *surface design*; this round explores *where on the build page*-the surface lives. Orthogonal axis: any style (A/B/C) can sit at any placement.--## Frame (placement)--The surface only renders in **debug/browser mode** (`showBrowser`), inside the vertical-`Splitter`. Candidate regions and their occupants:-- **Left pane** = code (left half) + infinite canvas (right half). Bottom-left corner holds the- FlowRenderer controls (FitView/Lock/GlobalCollapse); top is overlaid by `WorkflowHeader`- (`absolute left-6 top-8 h-20`) and header-anchored sub-panels (`top-[8.5rem]`).-- **Right pane** (`skyvern-split-right`) = live browser + footer (copilot/power/reload); far- right edge holds the run-detail **timeline rail** (`w-[5rem]`, expandable, `z-[15]`).--**Deciding constraint:** a placement must not block the live browser, the canvas controls, or-the run-detail timeline, and should not fight the header/sub-panels. So placements are-pointer-events-transparent floating bands (only the dock itself captures clicks).--## Inspiration (Mobbin — web, node-canvas editors)-- **WRITER — Blueprints**: execution log as a full-width **bottom** panel (node rows + status +- Trace + duration). [link](https://mobbin.com/screens/2368f17d-4fcc-4590-912d-08cb39245dab)-- **n8n — Executions**: run history as a **left side panel** (status + duration, auto-refresh),- entered via a top **Editor | Executions** tab.- [link](https://mobbin.com/screens/d3af4413-04f9-4534-8aa4-01916d2b4910)-- **StackAI — Version history / Run Details**: **right side panel** co-located with the run- (status, timestamps, duration, tokens).- [link](https://mobbin.com/screens/cffa32bb-fa1e-43fa-913b-27001b35f395) ·- [link](https://mobbin.com/screens/bb0174f4-60aa-4e30-ac5f-73679b160f38)-- **Databricks** (round 1): top **run-selector** + bottom **event log** in one editor.--## Placement concepts--- **P1 — Bottom dock (baseline, current):** floating bottom-center over the canvas/code pane.- Lineage: WRITER bottom log + StackAI bottom toolbar. + canvas stays clear, controls reachable.- − easy to miss; far from where the eye starts.-- **P2 — Top rail (under header):** pinned top-center over the canvas/code pane, just below the- header. Lineage: n8n Editor|Executions top tab + Databricks top run-selector. + high- discoverability, reads top-down, near run controls. − competes with header sub-panels when one- is open; popover opens downward over the canvas.-- **P3 — Browser-pane header (co-located with run viewing):** pinned to the top of the right- (live-browser) pane, so picking a recent run sits beside viewing that run + its detail- timeline. Lineage: StackAI right-side run history/details. + strongest pick→view coherence- (directly answers the ticket's "reconcile with run viewing"). − eats a strip of browser space;- must clear the far-right timeline rail.--**Recommendation:** **P3 (browser-pane header)** for coherence — it puts "which run" right next-to "this run's browser + timeline," which is the actual debugging loop — with **P1** as the-low-risk default if we want the canvas untouched. P2 is the most discoverable but the most-crowded.--**Decision (2026-06-16):** per the established pattern, build all three behind a **placement-toggle** (orthogonal to the A/B/C style toggle; persisted via localStorage + `?activityPlacement=`-URL seed; default `bottom`) so they can be compared live in the PR preview.--### Gap-ledger addition-- No shared **floating-dock anchor** convention for build-canvas overlays (each overlay- re-implements `absolute … pointer-events-none [&>*]:pointer-events-auto`). Minor; noted on- [[SKY-11082]] rather than a new ticket.--## Round 2 decision — style converges to one "Activity Timeline" (2026-06-16)--Requester: "I like C the most, and B for its compression and all the events in one place."-→ **Merge B + C into a single default surface** and drop A (and the A/B/C style toggle):--**Activity Timeline** — compact two-row dock:-- Row 1: a **run selector** ("Run: ✓ Login · 2m ago ▾") + a compression summary ("N runs ·- last ✓ 2m ago"). The selector opens **B's consolidated, scannable event list**- (`RecentActivityList`) — all events in one place.-- Row 2: a **persistent segmented timeline bar** (C) — proportional, status-colored; hover→tip,- click a segment → jump to that run.--This keeps C's timeline metaphor + direct segment navigation, with B's compression (compact-collapsed form) and single consolidated list. Built behind the **placement toggle** (bottom /-top / right) so location can still be compared in the PR preview.--## Round 3 decision — placement locked to browser-pane header (2026-06-16)--Requester locked the **browser-pane header (right)** placement. Removed the placement toggle +-store and the bottom/top anchors; `RecentActivity` now mounts unconditionally at the top of the-live-browser pane (`popoverSide="bottom"`). Rendered visual proof (mock runs) below — produced-via a throwaway standalone preview page, not committed.----## Round 4 — re-added the 3-way placement toggle (2026-06-16)--Reverted the round-3 lock: requester wants to compare bottom / top / right **by eye** in the-preview rather than commit blind. Restored `RecentActivityPlacementStore` +-`RecentActivityPlacementToggle` and the three Workspace anchors (default `bottom`,-`?activityPlacement=` URL seed). Decide, then re-lock.----## Round 5 — drop floating, integrate the run selector (2026-06-16)--Real screenshots showed every floating placement collides with canvas/browser content and-duplicates the existing right-side run-detail panel (`DebuggerRunTimeline`: Actions/Steps/Credits-+ block/action timeline) — the overlap the ticket warned about. So: **remove the floating dock,-segmented bar, and placement toggle**, and make recent-activity a compact **run selector** that-opens the consolidated list. Two integration points behind a small mode toggle (default `panel`,-`?activityMode=panel|toolbar`):-- **panel:** selector at the top of the right run-detail panel (`DebuggerRun`) — co-located with- viewing; picking a run drives the panel below.-- **toolbar:** selector as a dropdown in the editor header toolbar (`EditorActionToolbar`).-Removed: `RecentActivity`, `RecentActivityTimeline`, `RecentActivitySegmentBar`,-`RecentActivityPlacementToggle`, `RecentActivityPlacementStore`. Added: `RecentActivityRunSelector`,-`RecentActivityIntegrationStore`, `RecentActivityIntegrationToggle`.--## Round 6 — timeline strip below the browser (2026-06-16)--Panel/toolbar selectors still read as bolted-on. Requester idea: dock the timeline as a-horizontal strip **below the live browser session window** — the browser is what you're-watching, so a run history beneath it is contextual (WRITER bottom log / Databricks event-log-lineage). Added as a third integration option (`browser`) on the mode toggle:-`RecentActivityStrip` — a scannable horizontal row of run chips (oldest→newest, auto-scrolled to-newest, click→jump) + an "N runs ▾" button opening the consolidated list. Mounted between the-browser stream and its footer in both VNC and CDP panels. Compare via the Panel/Toolbar/Browser-toggle (`?activityMode=panel|toolbar|browser`).--## Round 7 — placement locked to the run-detail panel; preview toggle removed (final, 2026-06-29)--Compared the three integrations live in the PR preview and **locked `panel`**: the run-selector mounts at the top of the right run-detail panel (`DebuggerRun`), co-located with-viewing a run, which reconciles the history surface with run viewing (the ticket's open-question) without a second overlapping surface. Removed the exploration scaffolding —-`RecentActivityIntegrationStore` (+test), `RecentActivityIntegrationToggle`, and-`RecentActivityStrip` — plus the toolbar/browser wirings in `WorkflowHeader`/`Workspace`.-`RecentActivityRunSelector` mounts unconditionally (it self-hides until the debug session has-≥1 run) and its now-single-use `variant`/`popoverSide` props were dropped. Run rows in the-consolidated list are disabled while a workflow run is in progress (navigation is a no-op-then), covered by `RecentActivityList.test.tsx`. Dead exports `SEGMENT_FILL` and-`getRunTooltip` (orphaned when the segment bar / strip were cut) were removed.diff --git a/docs/design/activity-timeline-placements.png b/docs/design/activity-timeline-placements.pngdeleted file mode 100644index a5f1b6086..000000000Binary files a/docs/design/activity-timeline-placements.png and /dev/null differdiff --git a/docs/design/activity-timeline-preview.png b/docs/design/activity-timeline-preview.pngdeleted file mode 100644index 2b4b64f42..000000000Binary files a/docs/design/activity-timeline-preview.png and /dev/null differdiff --git a/docs/media-cdn-manifest.md b/docs/media-cdn-manifest.mddeleted file mode 100644index 5f6973f03..000000000--- a/docs/media-cdn-manifest.md+++ /dev/null@@ -1,140 +0,0 @@-# Documentation media CDN manifest--Skyvern's large public-documentation media is stored in-[`Skyvern-AI/skyvern-docs-media`](https://github.com/Skyvern-AI/skyvern-docs-media) and served through-jsDelivr. References must use content-addressed filenames and the full immutable media commit-`383c5fb7f1c10a89e57a34e503b05b7ee4e80a6a`.--## Published assets--### `docs/images/skyvern-agent-local.mp4`--- Original: SHA-256 `6da40c69458f7ea9dd3bc697d6dca7f790b632a4ef165eb68ee70189c7ec756e`, 21,492,235 bytes-- Hosted path: `media/quickstart/56f8b625b7ac6324f44e76fb13c14ebd546350ee97aa3ba76bceec62f579dc15-skyvern-agent-local.mp4`-- Hosted: SHA-256 `56f8b625b7ac6324f44e76fb13c14ebd546350ee97aa3ba76bceec62f579dc15`, 5,475,920 bytes, `video/mp4`-- URL: https://cdn.jsdelivr.net/gh/Skyvern-AI/skyvern-docs-media@383c5fb7f1c10a89e57a34e503b05b7ee4e80a6a/media/quickstart/56f8b625b7ac6324f44e76fb13c14ebd546350ee97aa3ba76bceec62f579dc15-skyvern-agent-local.mp4-- Re-encode: H.264 CRF 18 with `yuv420p` and fast-start metadata. Duration (36.333333 seconds),- resolution (3840x1852), frame rate (15 fps), and absence of audio are preserved. Full-file quality- comparison: SSIM 0.999451, PSNR 56.692168 dB, and sampled VMAF 97.053543.--### `docs/images/zapier-forwarding-setup.mp4`--- Original: SHA-256 `39a279a5540ad3aeed3717a9fb307c4fb2d0eac61a462b91320b44f54a4bd60b`, 14,363,999 bytes-- Hosted path: `media/authentication/39a279a5540ad3aeed3717a9fb307c4fb2d0eac61a462b91320b44f54a4bd60b-zapier-forwarding-setup.mp4`-- Hosted: SHA-256 `39a279a5540ad3aeed3717a9fb307c4fb2d0eac61a462b91320b44f54a4bd60b`, 14,363,999 bytes, `video/mp4`-- URL: https://cdn.jsdelivr.net/gh/Skyvern-AI/skyvern-docs-media@383c5fb7f1c10a89e57a34e503b05b7ee4e80a6a/media/authentication/39a279a5540ad3aeed3717a9fb307c4fb2d0eac61a462b91320b44f54a4bd60b-zapier-forwarding-setup.mp4-- Re-encode: none; copied byte-for-byte.--### `docs/images/skyvern-init.gif`--- Original: SHA-256 `8d8dc46d95f62f20c9ee6e433a1dcca0e2bb7b074fe93a2d56bc4fcd7ac85d4a`, 9,345,750 bytes-- Hosted path: `media/quickstart/8d8dc46d95f62f20c9ee6e433a1dcca0e2bb7b074fe93a2d56bc4fcd7ac85d4a-skyvern-init.gif`-- Hosted: SHA-256 `8d8dc46d95f62f20c9ee6e433a1dcca0e2bb7b074fe93a2d56bc4fcd7ac85d4a`, 9,345,750 bytes, `image/gif`-- URL: https://cdn.jsdelivr.net/gh/Skyvern-AI/skyvern-docs-media@383c5fb7f1c10a89e57a34e503b05b7ee4e80a6a/media/quickstart/8d8dc46d95f62f20c9ee6e433a1dcca0e2bb7b074fe93a2d56bc4fcd7ac85d4a-skyvern-init.gif-- Re-encode: none; copied byte-for-byte.--### `docs/images/zapier/zap1.mp4`--- Original: SHA-256 `06b4c3ca491de54d8f46625f29bc7e220a41f26690a7b83b498e5656b31d12b2`, 4,889,331 bytes-- Hosted path: `media/integrations/zapier/06b4c3ca491de54d8f46625f29bc7e220a41f26690a7b83b498e5656b31d12b2-zap1.mp4`-- Hosted: SHA-256 `06b4c3ca491de54d8f46625f29bc7e220a41f26690a7b83b498e5656b31d12b2`, 4,889,331 bytes, `video/mp4`-- URL: https://cdn.jsdelivr.net/gh/Skyvern-AI/skyvern-docs-media@383c5fb7f1c10a89e57a34e503b05b7ee4e80a6a/media/integrations/zapier/06b4c3ca491de54d8f46625f29bc7e220a41f26690a7b83b498e5656b31d12b2-zap1.mp4-- Re-encode: none; copied byte-for-byte.--### `docs/images/zapier/zap2.mp4`--- Original: SHA-256 `8feb209615ef1199d75cb94844f5ad9b6ec0d31c5e5a91344615810801bd522b`, 3,260,575 bytes-- Hosted path: `media/integrations/zapier/8feb209615ef1199d75cb94844f5ad9b6ec0d31c5e5a91344615810801bd522b-zap2.mp4`-- Hosted: SHA-256 `8feb209615ef1199d75cb94844f5ad9b6ec0d31c5e5a91344615810801bd522b`, 3,260,575 bytes, `video/mp4`-- URL: https://cdn.jsdelivr.net/gh/Skyvern-AI/skyvern-docs-media@383c5fb7f1c10a89e57a34e503b05b7ee4e80a6a/media/integrations/zapier/8feb209615ef1199d75cb94844f5ad9b6ec0d31c5e5a91344615810801bd522b-zap2.mp4-- Re-encode: none; copied byte-for-byte.--### `docs/images/n8n/n8n1.mp4`--- Original: SHA-256 `2248b0eb8f8dfadd4196478712f8fb65de1ef2544cf6f513d9c7ec3c9f4f119e`, 4,468,871 bytes-- Hosted path: `media/integrations/n8n/2248b0eb8f8dfadd4196478712f8fb65de1ef2544cf6f513d9c7ec3c9f4f119e-n8n1.mp4`-- Hosted: SHA-256 `2248b0eb8f8dfadd4196478712f8fb65de1ef2544cf6f513d9c7ec3c9f4f119e`, 4,468,871 bytes, `video/mp4`-- URL: https://cdn.jsdelivr.net/gh/Skyvern-AI/skyvern-docs-media@383c5fb7f1c10a89e57a34e503b05b7ee4e80a6a/media/integrations/n8n/2248b0eb8f8dfadd4196478712f8fb65de1ef2544cf6f513d9c7ec3c9f4f119e-n8n1.mp4-- Re-encode: none; copied byte-for-byte.--### `docs/images/make/make1.mp4`--- Original: SHA-256 `4a6af7c02a875f43863583e5929368f0a1783a8f2457690b6a2b359343197627`, 4,249,598 bytes-- Hosted path: `media/integrations/make/4a6af7c02a875f43863583e5929368f0a1783a8f2457690b6a2b359343197627-make1.mp4`-- Hosted: SHA-256 `4a6af7c02a875f43863583e5929368f0a1783a8f2457690b6a2b359343197627`, 4,249,598 bytes, `video/mp4`-- URL: https://cdn.jsdelivr.net/gh/Skyvern-AI/skyvern-docs-media@383c5fb7f1c10a89e57a34e503b05b7ee4e80a6a/media/integrations/make/4a6af7c02a875f43863583e5929368f0a1783a8f2457690b6a2b359343197627-make1.mp4-- Re-encode: none; copied byte-for-byte.--### `docs/images/make/make2.mp4`--- Original: SHA-256 `4220a02d47007e40468b5d2356b3ed8c53603e89532b88835572aded35e582dd`, 1,875,634 bytes-- Hosted path: `media/integrations/make/4220a02d47007e40468b5d2356b3ed8c53603e89532b88835572aded35e582dd-make2.mp4`-- Hosted: SHA-256 `4220a02d47007e40468b5d2356b3ed8c53603e89532b88835572aded35e582dd`, 1,875,634 bytes, `video/mp4`-- URL: https://cdn.jsdelivr.net/gh/Skyvern-AI/skyvern-docs-media@383c5fb7f1c10a89e57a34e503b05b7ee4e80a6a/media/integrations/make/4220a02d47007e40468b5d2356b3ed8c53603e89532b88835572aded35e582dd-make2.mp4-- Re-encode: none; copied byte-for-byte.--### `docs/images/browser-session-create.mp4`--- Original: SHA-256 `3c68f88cf30fe223cc7dfe2edd90e03054377d702b1c619e65cfef3c064efe41`, 3,735,886 bytes-- Hosted path: `media/cookbooks/healthcare/3c68f88cf30fe223cc7dfe2edd90e03054377d702b1c619e65cfef3c064efe41-browser-session-create.mp4`-- Hosted: SHA-256 `3c68f88cf30fe223cc7dfe2edd90e03054377d702b1c619e65cfef3c064efe41`, 3,735,886 bytes, `video/mp4`-- URL: https://cdn.jsdelivr.net/gh/Skyvern-AI/skyvern-docs-media@383c5fb7f1c10a89e57a34e503b05b7ee4e80a6a/media/cookbooks/healthcare/3c68f88cf30fe223cc7dfe2edd90e03054377d702b1c619e65cfef3c064efe41-browser-session-create.mp4-- Re-encode: none; copied byte-for-byte.--### `docs/images/navigation-error-messages.mp4`--- Original: SHA-256 `dd90d50875e60e6a99319157683bfc61a227db9772018eccde3c31ff039f2f3e`, 3,581,629 bytes-- Hosted path: `media/cookbooks/healthcare/dd90d50875e60e6a99319157683bfc61a227db9772018eccde3c31ff039f2f3e-navigation-error-messages.mp4`-- Hosted: SHA-256 `dd90d50875e60e6a99319157683bfc61a227db9772018eccde3c31ff039f2f3e`, 3,581,629 bytes, `video/mp4`-- URL: https://cdn.jsdelivr.net/gh/Skyvern-AI/skyvern-docs-media@383c5fb7f1c10a89e57a34e503b05b7ee4e80a6a/media/cookbooks/healthcare/dd90d50875e60e6a99319157683bfc61a227db9772018eccde3c31ff039f2f3e-navigation-error-messages.mp4-- Re-encode: none; copied byte-for-byte.--### `docs/images/workflow-start.mp4`--- Original: SHA-256 `1dcbdffcdff31b8a67458ed3b64fe1d17f17f6b3c10b18bfbde92a75391daab0`, 3,177,949 bytes-- Hosted path: `media/cookbooks/healthcare/1dcbdffcdff31b8a67458ed3b64fe1d17f17f6b3c10b18bfbde92a75391daab0-workflow-start.mp4`-- Hosted: SHA-256 `1dcbdffcdff31b8a67458ed3b64fe1d17f17f6b3c10b18bfbde92a75391daab0`, 3,177,949 bytes, `video/mp4`-- URL: https://cdn.jsdelivr.net/gh/Skyvern-AI/skyvern-docs-media@383c5fb7f1c10a89e57a34e503b05b7ee4e80a6a/media/cookbooks/healthcare/1dcbdffcdff31b8a67458ed3b64fe1d17f17f6b3c10b18bfbde92a75391daab0-workflow-start.mp4-- Re-encode: none; copied byte-for-byte.--### `docs/images/laminar-keys.mp4`--- Original: SHA-256 `f2d0b8714b6e6ec3b0b50d9a6e3384897a1d093f47c0542c9d6e1dc69b3fddee`, 1,335,413 bytes-- Hosted path: `media/observability/f2d0b8714b6e6ec3b0b50d9a6e3384897a1d093f47c0542c9d6e1dc69b3fddee-laminar-keys.mp4`-- Hosted: SHA-256 `f2d0b8714b6e6ec3b0b50d9a6e3384897a1d093f47c0542c9d6e1dc69b3fddee`, 1,335,413 bytes, `video/mp4`-- URL: https://cdn.jsdelivr.net/gh/Skyvern-AI/skyvern-docs-media@383c5fb7f1c10a89e57a34e503b05b7ee4e80a6a/media/observability/f2d0b8714b6e6ec3b0b50d9a6e3384897a1d093f47c0542c9d6e1dc69b3fddee-laminar-keys.mp4-- Re-encode: none; copied byte-for-byte.--## Verification--On 2026-07-24, anonymous full downloads of all 12 URLs matched the hosted SHA-256 and byte length.-Responses used the declared `video/mp4` or `image/gif` type and-`Cache-Control: public, max-age=31536000, s-maxage=31536000, immutable`. Each MP4 returned a 1,024-byte-`206 Partial Content` response for `Range: bytes=0-1023`, with the expected `Content-Range` and-`Accept-Ranges: bytes`.--A clean Chromium 145 session decoded all 11 MP4s with nonzero dimensions and finite duration, reported no-media errors, and advanced playback. The GIF decoded at 1376x1080.--## Publishing updates--1. Review jsDelivr's [GitHub endpoint](https://github.com/jsdelivr/jsdelivr#github),- [package restrictions](https://github.com/jsdelivr/jsdelivr#restrictions), and- [caching contract](https://github.com/jsdelivr/jsdelivr#caching). Keep the repository below 150 MB and- every file below 20 MB.-2. Work in a temporary clone of `Skyvern-AI/skyvern-docs-media`, never in the documentation repository.- Calculate the source SHA-256 and exact byte length before transforming anything.-3. If an MP4 exceeds 20 MB, encode an H.264 `yuv420p` copy with fast-start metadata. Compare `ffprobe`- duration, dimensions, frame rate, and audio streams; run SSIM/PSNR or VMAF; inspect representative- frames; and test browser playback.-4. Put the asset in a logical `media/` subdirectory. Prefix its filename with the full hosted SHA-256.- Commit and push the media. Never modify or remove a published origin commit.-5. Build the URL as- `https://cdn.jsdelivr.net/gh/Skyvern-AI/skyvern-docs-media@<full-commit-sha>/<hosted-path>`. Never use a- branch, tag, shortened commit, version range, or latest alias.-6. Before changing docs, anonymously verify the full response's checksum, length, MIME type, and one-year- immutable cache policy. Verify `206` byte-range behavior for MP4 and decode every asset in clean- Chromium.-7. Add the source and hosted metadata to this manifest, update only the consumer URL, and preserve all- video attributes, source types, styles, and image alt text. Delete a Git copy only after its CDN URL- passes every check.diff --git a/docs/plans/2026-07-13-totp-uri-placeholder-fix.md b/docs/plans/2026-07-13-totp-uri-placeholder-fix.mddeleted file mode 100644index d65cacdfc..000000000--- a/docs/plans/2026-07-13-totp-uri-placeholder-fix.md+++ /dev/null@@ -1,28 +0,0 @@-# TOTP issuer mismatch and placeholder fail-closed plan--## Scope--Fix standard `otpauth://` TOTP URIs whose label issuer differs from the `issuer` query parameter, while retaining strict validation of the secret and runtime configuration. Prevent unresolved TOTP placeholders from reaching browser input paths.--## Implementation--1. Add regression tests in `tests/unit/test_parse_totp_secret.py` for raw and percent-encoded issuer mismatches, non-default TOTP settings, and malformed configurations that must remain rejected.-2. Update `skyvern/forge/sdk/services/credentials.py` so strict `pyotp.parse_uri()` validation remains the default and only the issuer-mismatch error gets a validation-only retry without issuer query metadata.-3. Add handler and generated-script tests proving missing TOTP material raises a typed failure and browser typing is never called across DOM, selectable, CUA, and autocomplete paths.-4. Update `skyvern/webeye/actions/handler.py` and `skyvern/core/script_generations/skyvern_page.py` so every input path resolves a TOTP marker before branching or typing.--## Acceptance criteria--- Mismatched label/query issuers generate the correct code while preserving algorithm, digit count, and period.-- Matching-issuer URIs and raw Base32 secrets behave unchanged.-- Missing or invalid secrets, unsupported algorithms or digit counts, malformed periods, unknown parameters, and HOTP URIs remain rejected.-- No unresolved `placeholder_*_totp` value is passed to a browser typing method.-- Failures do not expose the URI, secret, placeholder, or generated code.--## Verification--- Run targeted parser and handler tests during TDD.-- Compile and lint every modified Python file.-- Run the broader credential/TOTP regression suites and all unit tests.-- Run pre-commit on the changed files and `git diff --check`.-- Obtain independent parser/security and action-failure reviews, then complete a separate QA pass.diff --git a/docs/plans/2026-08-03-sky-13291-vendor-live-view-metadata.md b/docs/plans/2026-08-03-sky-13291-vendor-live-view-metadata.mddeleted file mode 100644index 0fd5200ef..000000000--- a/docs/plans/2026-08-03-sky-13291-vendor-live-view-metadata.md+++ /dev/null@@ -1,72 +0,0 @@-# SKY-13291 vendor live-view metadata persistence--## Scope--This PR is the persistence prerequisite for vendor live view. It captures the credential-bearing-live-view URL and a provider-neutral protocol value at the provider response, carries them only in-the vendor provisioning activity, and stores them only in `browser_session_infra`.--It does not add a viewer, router relay, API field, CSP rule, or frontend code. Those paths remain-blocked on a documented raw RFB endpoint for headful Anchor sessions and a Browser Use strategy-that does not send vendor HTML or JavaScript to customers.--## Implementation--1. Add nullable `live_view_url` and `live_view_protocol` columns to- `browser_session_infra` with one additive Alembic migration.-2. Extend the cloud-only infrastructure repository to atomically attach the provider session id- and live-view metadata after provisioning. Keep both sensitive fields out of dataclass reprs.-3. Capture Anchor's live-view URL from both supported provisioning transports. Classify current- headful sessions as `rfb` and future explicitly headless sessions as `cdp_screencast`.-4. Capture Browser Use's `live_url` and classify it as `hosted_html`.-5. Carry the metadata in the activity-local `VendorBrowser` only. Do not add it to- `VendorProvisionResult`, `PersistentBrowserSession`, or any API schema.--## Acceptance criteria--- A successful provider provision stores its live-view URL and generic protocol in- `browser_session_infra` together with the provider session id.-- Missing live-view metadata remains nullable and does not make an otherwise usable browser fail.-- The credential does not appear in object repr/string output, structured logs, Temporal activity- payloads, persistent browser-session models, OpenAPI schemas, or client code.-- Existing provider-identity non-disclosure tests remain green.-- The migration graph has exactly one head.--## Verification--```bash-uv run pytest tests/cloud/test_anchor_session_provisioner.py \- tests/cloud/test_browser_session_infra.py \- tests/cloud/test_pbs_vendor_session_lifecycle.py \- tests/cloud/test_vendor_identity_nondisclosure.py \- tests/unit/webeye/test_browser_session_response.py -q-uv run pytest tests/cloud/test_migration_chain.py -q-uv run alembic heads-uv run python -m py_compile <each modified Python file>-uv run ruff check <modified Python files>-uv run pre-commit run --files <modified files>-```--Each security assertion is mutation-checked by temporarily removing its production guard, running-the focused test to observe failure, restoring the guard, and rerunning the test to green.--## Escalation: Size Limits--**What I tried**: Implemented the human-approved persistence-only slice with the six requested-mutation-proven confinement checks, then fixed three credential-safety findings from independent-review: SQL bound-parameter redaction, provider traceback-local scrubbing, and cancellation-safe-credential cleanup.--**What blocked me**: The resulting code diff is 617 changed lines excluding this plan and the new-migration: 584 additions and 33 deletions across five production and four test files. The-repository stop limit is 500 changed lines. The split is 197 production lines and 420 test lines;-the requested security and mutation coverage accounts for most of the excess. The change remains-within the file-count and subsystem limits.--**Decision needed**: A human must either approve a size-limit exception for this security-focused-slice or direct a specific reduction/split. No commit, push, or PR will be created without that-decision.--**My recommendation**: Approve the exception for this slice. Splitting the confinement tests from-the credential persistence they protect would make either PR independently unsafe, and deleting-tests would weaken the explicit mutation-proof requirement.diff --git a/docs/plans/2026-08-06-sky-13291-vendor-live-view-cdp-screencast.md b/docs/plans/2026-08-06-sky-13291-vendor-live-view-cdp-screencast.mddeleted file mode 100644index 7ff4c66de..000000000--- a/docs/plans/2026-08-06-sky-13291-vendor-live-view-cdp-screencast.md+++ /dev/null@@ -1,865 +0,0 @@-# Vendor Live View via Per-Session CDP Screencast (SKY-13291 slice 3) Implementation Plan--> **For agentic workers:** REQUIRED SUB-SKILL: Use superpowers:subagent-driven-development (recommended) or superpowers:executing-plans to implement this plan task-by-task. Steps use checkbox (`- [ ]`) syntax for tracking.--> **Amended after review — the run surfaces are out of scope, and the viewer connects differently.**-> Tasks 1–3 below plan a per-session gate on the task and workflow-run stream endpoints, backed by-> a fallback that resolves a run's browser through its session. Review established that both parts-> were wrong, and what shipped differs:->-> - **Run surfaces keep the deployment-global gate.** A run's worker already publishes frames for a-> remote-CDP browser (`CDPFramePublisher` → `StorageBase.get_streaming_file`), which is how those-> endpoints already serve an externally hosted session. Routing them per session bought nothing —-> and missed anyway, because a task inside a workflow leases its session under the *run's* id, so-> resolving by task id never matched. `verify.browser_session_id_for_runnable` was dropped.-> - **The viewer no longer reuses `get_browser_state`.** For an externally hosted session that adopts-> the browser through the Fetch-download factory, which would arm a second download interceptor on-> pages a run is driving and leave the connection in a cache a run later reads. Watching now goes-> through `get_observer_browser_state` / `release_observer_browser_state`: one uncached,-> interception-free connection per websocket, given back when the viewer leaves.->-> The per-session gate on the browser-session endpoint, the `stream_transport` field, the frontend-> selection, and the capability flip are as planned.--**Goal:** Make live-stream transport a per-session decision instead of the deployment-global `BROWSER_STREAMING_MODE`, so external-provider ("vendor") browser sessions stream via CDP `Page.startScreencast` through the session router while first-party sessions keep VNC.--**Architecture:** A new `AgentFunction.resolve_stream_transport` hook answers `"vnc"` or `"cdp"` per session (OSS base returns the global setting; the cloud override answers `"cdp"` for vendor-held rows from `browser_session_infra`). The four backend serving-path gates that read `settings.BROWSER_STREAMING_MODE` switch to the hook, the CDP screencast path learns to resolve a task/workflow-run's browser state through `PERSISTENT_SESSIONS_MANAGER` (remote `connect_over_cdp` on the routed address) when the in-process `BROWSER_MANAGER` has none, `BrowserSessionResponse` grows a `stream_transport` field, and the frontend's 7 stream-selection call sites consult the session's transport with the global mode as fallback. The final, separately-revertable step flips `live_view=True` in the vendor capability table, which simultaneously permits routing `needs_live_view` sessions to vendors and (via the existing `supports_live_view` override, which already delegates to the capability table) makes `vnc_streaming_supported` true for vendor sessions.--**Tech Stack:** Python/FastAPI (skyvern OSS + cloud overlay), Playwright CDP sessions, React/TypeScript + react-query (skyvern-frontend), pytest (`@pytest.mark.asyncio`), vitest.--**Context (verified 2026-08-06):** Both vendors are spike-proven for `Page.startScreencast` on the shared driver connection (41–53 fps, zero driver interference, second concurrent CDP connection accepted). See the SKY-13291 Linear comments for numbers. The viewer stack (screencast loop, CDP input channel, `InteractiveStreamView`) already exists from OSS local mode.--## Global Constraints--- **OSS boundary:** never import `cloud.*` from `skyvern/`. All cloud-specific logic goes through the `AgentFunction` override in `cloud/agent_functions.py`.-- **OSS sync / non-disclosure:** `skyvern/`, `tests/unit*/` and `docs/` are all synced to the public repo (see `.github/sync.yml` for the full list — this document is inside that surface). No provider name may appear anywhere in it — say "external provider". Provider-specific logic and tests live in `cloud/` and `tests/cloud/`.-- Line length 120; `structlog` (`LOG = structlog.get_logger()`), imports at top of file.-- Comments: only for non-obvious constraints; no change-log or task-reference comments.-- Tests: extend existing `test_X` files for module X; check `tests/unit/conftest.py` fixtures before writing fakes; no exact-prompt or mock-call-count assertions.-- Branch naming: `{username}/sky-13291-{short-description}`. Commits use conventional messages. No migrations are needed (no schema changes — `stream_transport` is computed, not stored).-- PR bodies use `.github/pull_request_template.md` (Problem / Solution / How Has This Been Tested?); validate with `python3 .github/scripts/validate_pr_body.py .github/pull_request_template.md <body-file>`.-- Validation floor per modified Python file: `uv run python -m py_compile <file>`; frontend: `cd skyvern-frontend && npx tsc --noEmit` and `npx eslint <files>`.-- Suggested PR split: Tasks 1–5 (backend), Tasks 6–7 (frontend), Task 8 (capability flip — its own PR, the rollout switch).-------### Task 1: `resolve_stream_transport` AgentFunction + resolution helpers--**Files:**-- Modify: `skyvern/forge/agent_functions.py` (next to `supports_live_view`, ~line 917)-- Modify: `skyvern/forge/sdk/routes/streaming/verify.py` (module level, after `verify_browser_session`)-- Test: `tests/unit/forge/sdk/routes/streaming/test_stream_transport.py` (new)--**Interfaces:**-- Consumes: `settings.BROWSER_STREAMING_MODE` (str, `"vnc"` default), `app.AGENT_FUNCTION`, `app.PERSISTENT_SESSIONS_MANAGER.get_session_by_runnable_id(organization_id=..., runnable_id=...)`.-- Produces (later tasks rely on these exact names):- - `AgentFunction.resolve_stream_transport(*, browser_session_id: str | None, organization_id: str | None) -> str`- - `verify.stream_transport(browser_session_id: str | None, organization_id: str) -> str` (never raises)- - `verify.browser_session_id_for_runnable(runnable_id: str, organization_id: str) -> str | None` (never raises)--- [ ] **Step 1: Write the failing tests**--Create `tests/unit/forge/sdk/routes/streaming/test_stream_transport.py`:--```python-from unittest.mock import AsyncMock--import pytest--from skyvern.forge import app-from skyvern.forge.sdk.routes.streaming import verify---async def test_stream_transport_delegates_to_agent_function(monkeypatch: pytest.MonkeyPatch) -> None:- monkeypatch.setattr(app, "AGENT_FUNCTION", AsyncMock(resolve_stream_transport=AsyncMock(return_value="cdp")))-- assert await verify.stream_transport("pbs_1", "org_1") == "cdp"- app.AGENT_FUNCTION.resolve_stream_transport.assert_awaited_once_with(- browser_session_id="pbs_1", organization_id="org_1"- )---async def test_stream_transport_falls_back_to_setting_on_error(monkeypatch: pytest.MonkeyPatch) -> None:- monkeypatch.setattr(- app, "AGENT_FUNCTION", AsyncMock(resolve_stream_transport=AsyncMock(side_effect=RuntimeError("db down")))- )- monkeypatch.setattr(verify.settings, "BROWSER_STREAMING_MODE", "vnc")-- assert await verify.stream_transport("pbs_1", "org_1") == "vnc"---async def test_base_agent_function_returns_deployment_setting(monkeypatch: pytest.MonkeyPatch) -> None:- from skyvern.forge.agent_functions import AgentFunction-- monkeypatch.setattr(verify.settings, "BROWSER_STREAMING_MODE", "cdp")-- assert await AgentFunction().resolve_stream_transport(browser_session_id=None, organization_id=None) == "cdp"---async def test_browser_session_id_for_runnable_resolves_and_swallows(monkeypatch: pytest.MonkeyPatch) -> None:- session = AsyncMock()- session.persistent_browser_session_id = "pbs_9"- manager = AsyncMock(get_session_by_runnable_id=AsyncMock(return_value=session))- monkeypatch.setattr(app, "PERSISTENT_SESSIONS_MANAGER", manager)-- assert await verify.browser_session_id_for_runnable("wr_1", "org_1") == "pbs_9"-- manager.get_session_by_runnable_id.side_effect = RuntimeError("db down")- assert await verify.browser_session_id_for_runnable("wr_1", "org_1") is None-```--If this directory's existing tests use a different async marker or app-patching fixture (see `test_verify_browser_session.py`), match that style instead.--- [ ] **Step 2: Run tests to verify they fail**--Run: `uv run pytest tests/unit/forge/sdk/routes/streaming/test_stream_transport.py -v`-Expected: FAIL — `AttributeError` / `ImportError` (no `resolve_stream_transport`, no `verify.stream_transport`).--- [ ] **Step 3: Implement the AgentFunction base**--In `skyvern/forge/agent_functions.py`, directly below `supports_live_view` (the file already imports `settings`):--```python- async def resolve_stream_transport(- self, *, browser_session_id: str | None, organization_id: str | None- ) -> str:- """Which live-view transport serves this session: "vnc" or "cdp".-- A self-hosted deployment streams every browser the same way, so the- deployment-wide setting decides.- """- return settings.BROWSER_STREAMING_MODE-```--- [ ] **Step 4: Implement the helpers in verify.py**--In `skyvern/forge/sdk/routes/streaming/verify.py` (module already imports `app`, `settings`, `LOG`):--```python-async def stream_transport(browser_session_id: str | None, organization_id: str) -> str:- try:- return await app.AGENT_FUNCTION.resolve_stream_transport(- browser_session_id=browser_session_id, organization_id=organization_id- )- except Exception:- LOG.warning(- "Could not resolve the stream transport; using the deployment default",- browser_session_id=browser_session_id,- organization_id=organization_id,- exc_info=True,- )- return settings.BROWSER_STREAMING_MODE---async def browser_session_id_for_runnable(runnable_id: str, organization_id: str) -> str | None:- try:- session = await app.PERSISTENT_SESSIONS_MANAGER.get_session_by_runnable_id(- organization_id=organization_id, runnable_id=runnable_id- )- except Exception:- LOG.warning(- "Could not resolve a browser session for this runnable",- runnable_id=runnable_id,- organization_id=organization_id,- exc_info=True,- )- return None- return session.persistent_browser_session_id if session else None-```--- [ ] **Step 5: Run tests to verify they pass**--Run: `uv run pytest tests/unit/forge/sdk/routes/streaming/test_stream_transport.py -v`-Expected: PASS (4 tests).--- [ ] **Step 6: Compile check and commit**--```bash-uv run python -m py_compile skyvern/forge/agent_functions.py skyvern/forge/sdk/routes/streaming/verify.py-git add skyvern/forge/agent_functions.py skyvern/forge/sdk/routes/streaming/verify.py \- tests/unit/forge/sdk/routes/streaming/test_stream_transport.py-git commit -m "feat(SKY-13291): add per-session stream-transport resolution hook"-```-------### Task 2: Per-session transport at the four serving-path gates--**Files:**-- Modify: `skyvern/forge/sdk/routes/streaming/screenshot.py:68` (task stream), `:191` (workflow-run stream), `:339` (browser-session stream)-- Modify: `skyvern/forge/sdk/routes/streaming/verify.py:94` (address-readiness bypass inside `verify_browser_session`)-- Test: `tests/unit/forge/sdk/routes/streaming/test_verify_browser_session.py` (extend)--**Interfaces:**-- Consumes: `verify.stream_transport(...)` and `verify.browser_session_id_for_runnable(...)` from Task 1; existing `_local_screencast_for_task/_workflow_run/_browser_session` helpers in `screenshot.py`.-- Produces: no new symbols — behavior change only. After this task, a deployment with `BROWSER_STREAMING_MODE=vnc` serves the CDP screencast to any session whose `resolve_stream_transport` answers `"cdp"`, and vice versa.--- [ ] **Step 1: Write the failing test**--Append to `tests/unit/forge/sdk/routes/streaming/test_verify_browser_session.py` (match its existing fixtures for building a ready session; the essential assertions):--```python-async def test_verify_browser_session_cdp_transport_bypasses_missing_address(- monkeypatch: pytest.MonkeyPatch,-) -> None:- """A session whose transport is cdp verifies without a browser address; a vnc session still requires one."""- session = _make_session(is_browser_ready=False, browser_address=None) # reuse this file's session factory- manager = AsyncMock(get_session=AsyncMock(return_value=session))- monkeypatch.setattr(app, "PERSISTENT_SESSIONS_MANAGER", manager)- monkeypatch.setattr(app, "AGENT_FUNCTION", AsyncMock(resolve_stream_transport=AsyncMock(return_value="cdp")))- monkeypatch.setattr(verify.settings, "BROWSER_STREAMING_MODE", "vnc")-- result = await verify.verify_browser_session("pbs_1", "org_1")- assert result is not None- assert result.browser_address == ""-```--- [ ] **Step 2: Run test to verify it fails**--Run: `uv run pytest tests/unit/forge/sdk/routes/streaming/test_verify_browser_session.py -v -k cdp_transport`-Expected: FAIL — with the global set to `vnc`, the current code takes the address-lookup branch and returns `None`.--- [ ] **Step 3: Swap the verify.py bypass**--In `verify_browser_session` (`verify.py:94`), replace:--```python- if not browser_address:- if settings.BROWSER_STREAMING_MODE == "cdp":- browser_address = ""-```--with:--```python- if not browser_address:- if await stream_transport(browser_session_id, organization_id) == "cdp":- browser_address = ""-```--- [ ] **Step 4: Swap the three screenshot.py gates**--`screenshot.py` already imports from `verify`; extend that import with `stream_transport` and `browser_session_id_for_runnable`.--Task stream (line 68):--```python- task_session_id = await browser_session_id_for_runnable(task_id, organization_id)- if await stream_transport(task_session_id, organization_id) == "cdp":- await _local_screencast_for_task(websocket, task_id, organization_id)- return-```--Workflow-run stream (line 191):--```python- run_session_id = await browser_session_id_for_runnable(workflow_run_id, organization_id)- if await stream_transport(run_session_id, organization_id) == "cdp":- await _local_screencast_for_workflow_run(websocket, workflow_run_id, organization_id)- return-```--Browser-session stream (line 339):--```python- if await stream_transport(browser_session_id, organization_id) == "cdp":- await _local_screencast_for_browser_session(websocket, browser_session_id, organization_id)- return-```--Leave `settings.BROWSER_STREAMING_MODE` reads outside the serving path untouched (`runtime_config.py`, `forge_app.py`, `default_persistent_sessions_manager.py`, CLI).--- [ ] **Step 5: Run the streaming suites**--Run: `uv run pytest tests/unit/forge/sdk/routes/streaming/ tests/unit_tests/test_streaming_screencast.py -v`-Expected: PASS (new test green, no regressions — existing tests exercising the old gates keep passing because the base hook returns the same setting).--- [ ] **Step 6: Compile check and commit**--```bash-uv run python -m py_compile skyvern/forge/sdk/routes/streaming/screenshot.py skyvern/forge/sdk/routes/streaming/verify.py-git add skyvern/forge/sdk/routes/streaming/screenshot.py skyvern/forge/sdk/routes/streaming/verify.py \- tests/unit/forge/sdk/routes/streaming/test_verify_browser_session.py-git commit -m "feat(SKY-13291): gate stream endpoints on per-session transport"-```-------### Task 3: Screencast browser-state fallback through the persistent-session manager--**Files:**-- Modify: `skyvern/forge/sdk/routes/streaming/screencast.py:53-65` (`_resolve_browser_state`)-- Test: `tests/unit_tests/test_streaming_screencast.py` (extend — this is the module's existing test file)--**Interfaces:**-- Consumes: `app.BROWSER_MANAGER.get_for_workflow_run(entity_id)` / `.get_for_task(entity_id, workflow_run_id)` (in-process, sync); `app.PERSISTENT_SESSIONS_MANAGER.get_session_by_runnable_id(...)` and `.get_browser_state(session_id, organization_id)` (async, does `connect_over_cdp` on the routed address in cloud).-- Produces: `_resolve_browser_state` now returns a `BrowserState` for task/workflow-run entities whose browser lives behind a persistent session even when this process holds no in-process state (the cloud API server case). `wait_for_browser_state`, the screencast loop, and `cdp_input.py` (which imports `_resolve_working_page`) inherit the behavior unchanged.--**Why:** the run/task stream endpoints run on the API server; in cloud, the in-process `BROWSER_MANAGER` belongs to the temporal worker, so `get_for_workflow_run`/`get_for_task` return `None` there. The run's browser is still reachable: entity → persistent session → `get_browser_state` (remote CDP).--- [ ] **Step 1: Write the failing tests**--Append to `tests/unit_tests/test_streaming_screencast.py` (reuse its fake-browser-state helpers; `@pytest.mark.asyncio` is this file's convention):--```python-async def test_resolve_workflow_run_falls_back_to_persistent_session(monkeypatch: pytest.MonkeyPatch) -> None:- state = object()- session = AsyncMock()- session.persistent_browser_session_id = "pbs_7"- monkeypatch.setattr(app, "BROWSER_MANAGER", MagicMock(get_for_workflow_run=MagicMock(return_value=None)))- monkeypatch.setattr(- app,- "PERSISTENT_SESSIONS_MANAGER",- AsyncMock(- get_session_by_runnable_id=AsyncMock(return_value=session),- get_browser_state=AsyncMock(return_value=state),- ),- )-- resolved = await screencast._resolve_browser_state("wr_1", "workflow_run", organization_id="org_1")- assert resolved is state---async def test_resolve_workflow_run_without_session_or_org_returns_none(monkeypatch: pytest.MonkeyPatch) -> None:- monkeypatch.setattr(app, "BROWSER_MANAGER", MagicMock(get_for_workflow_run=MagicMock(return_value=None)))- monkeypatch.setattr(- app, "PERSISTENT_SESSIONS_MANAGER", AsyncMock(get_session_by_runnable_id=AsyncMock(return_value=None))- )-- assert await screencast._resolve_browser_state("wr_1", "workflow_run", organization_id="org_1") is None- assert await screencast._resolve_browser_state("wr_1", "workflow_run", organization_id=None) is None-```--Guard against `MagicMock` auto-attributes making assertions vacuous: assert identity (`is state`), not truthiness.--- [ ] **Step 2: Run tests to verify they fail**--Run: `uv run pytest tests/unit_tests/test_streaming_screencast.py -v -k persistent_session`-Expected: FAIL — current code returns whatever `BROWSER_MANAGER` returned (`None`) with no fallback.--- [ ] **Step 3: Implement the fallback**--Replace `_resolve_browser_state` in `screencast.py`:--```python-async def _resolve_browser_state(- entity_id: str,- entity_type: str,- workflow_run_id: str | None = None,- organization_id: str | None = None,-) -> BrowserState | None:- if entity_type == "workflow_run":- state = app.BROWSER_MANAGER.get_for_workflow_run(entity_id)- return state if state is not None else await _state_via_persistent_session(entity_id, organization_id)- if entity_type == "task":- state = app.BROWSER_MANAGER.get_for_task(entity_id, workflow_run_id)- return state if state is not None else await _state_via_persistent_session(entity_id, organization_id)- if entity_type == "browser_session":- return await app.PERSISTENT_SESSIONS_MANAGER.get_browser_state(entity_id, organization_id)- return None---async def _state_via_persistent_session(runnable_id: str, organization_id: str | None) -> BrowserState | None:- """The runnable's browser may live behind a persistent session another process owns; reach it remotely."""- if not organization_id:- return None- try:- session = await app.PERSISTENT_SESSIONS_MANAGER.get_session_by_runnable_id(- organization_id=organization_id, runnable_id=runnable_id- )- if not session:- return None- return await app.PERSISTENT_SESSIONS_MANAGER.get_browser_state(- session.persistent_browser_session_id, organization_id- )- except Exception:- LOG.debug("Could not resolve browser state via persistent session", runnable_id=runnable_id, exc_info=True)- return None-```--- [ ] **Step 4: Run the full screencast suite**--Run: `uv run pytest tests/unit_tests/test_streaming_screencast.py tests/unit/forge/sdk/routes/streaming/ -v`-Expected: PASS — including the SKY-11215 rebind tests (the fallback only engages when the manager returns `None`, so the resolve-each-poll behavior is unchanged).--- [ ] **Step 5: Compile check and commit**--```bash-uv run python -m py_compile skyvern/forge/sdk/routes/streaming/screencast.py-git add skyvern/forge/sdk/routes/streaming/screencast.py tests/unit_tests/test_streaming_screencast.py-git commit -m "feat(SKY-13291): resolve screencast browser state via persistent sessions when out-of-process"-```-------### Task 4: `stream_transport` on `BrowserSessionResponse`--**Files:**-- Modify: `skyvern/webeye/schemas.py` (field near `vnc_streaming_supported` at line 76; factory `from_browser_session` near line 176)-- Test: `tests/unit/webeye/test_browser_session_response.py` (extend)--**Interfaces:**-- Consumes: `app.AGENT_FUNCTION.resolve_stream_transport(...)` from Task 1.-- Produces: `BrowserSessionResponse.stream_transport: str | None` — the value the frontend reads in Task 6. `"vnc"` or `"cdp"`; `None` only for old serialized payloads.--**Note:** additive optional field ⇒ Fern SDK drift. Follow `cloud_docs/fern-sdk/` regeneration policy in the PR (drift check is advisory; call the field out in the PR body).--- [ ] **Step 1: Write the failing test**--Append to `tests/unit/webeye/test_browser_session_response.py` (reuse its session/factory fixtures):--```python-async def test_response_carries_stream_transport(monkeypatch: pytest.MonkeyPatch) -> None:- monkeypatch.setattr(- app, "AGENT_FUNCTION", AsyncMock(resolve_stream_transport=AsyncMock(return_value="cdp"), **_agent_function_defaults())- )- response = await BrowserSessionResponse.from_browser_session(_make_browser_session())- assert response.stream_transport == "cdp"-```--Where `_agent_function_defaults()` supplies whatever other AgentFunction methods this factory already calls (`resolve_browser_session_connect_url`, `supports_live_view`, …) — reuse the file's existing stubbing pattern rather than inventing a new one.--- [ ] **Step 2: Run test to verify it fails**--Run: `uv run pytest tests/unit/webeye/test_browser_session_response.py -v -k stream_transport`-Expected: FAIL — `stream_transport` not a field.--- [ ] **Step 3: Add the field and populate it**--Field (next to `vnc_streaming_supported`, line ~76):--```python- stream_transport: str | None = Field(- None,- description='Live-view transport for this session: "vnc" or "cdp".',- examples=["vnc", "cdp"],- )-```--In `from_browser_session`, alongside the `vnc_streaming_supported=` argument:--```python- stream_transport=await app.AGENT_FUNCTION.resolve_stream_transport(- browser_session_id=browser_session.persistent_browser_session_id,- organization_id=browser_session.organization_id,- ),-```--- [ ] **Step 4: Run tests to verify they pass**--Run: `uv run pytest tests/unit/webeye/test_browser_session_response.py tests/cloud/test_vendor_identity_nondisclosure.py -v`-Expected: PASS — including the non-disclosure suite: the field carries only a neutral transport word, never a vendor identity or URL.--- [ ] **Step 5: Compile check and commit**--```bash-uv run python -m py_compile skyvern/webeye/schemas.py-git add skyvern/webeye/schemas.py tests/unit/webeye/test_browser_session_response.py-git commit -m "feat(SKY-13291): expose per-session stream transport on browser session responses"-```-------### Task 5: Cloud override — vendor-held sessions stream via CDP--**Files:**-- Modify: `cloud/agent_functions.py` (next to `supports_live_view`, ~line 868)-- Test: `tests/cloud/test_stream_transport_resolution.py` (new — per-function cloud test files are this repo's pattern, cf. `test_browser_session_connect_url.py`)--**Interfaces:**-- Consumes: `cloud_db.browser_session_infra.get_browser_session_infra(persistent_browser_session_id) -> SessionInfra` (missing row ⇒ `FIRST_PARTY`; `SessionInfra.is_vendor: bool`). Both already imported/used in this module by `supports_live_view`.-- Produces: cloud `resolve_stream_transport` returning `"cdp"` for vendor rows, else `settings.BROWSER_STREAMING_MODE`.--- [ ] **Step 1: Write the failing tests**--Create `tests/cloud/test_stream_transport_resolution.py` (use the same infra-row fixtures as `tests/cloud/test_browser_session_infra.py` / the stubbing style of `test_browser_session_connect_url.py`):--```python-from unittest.mock import AsyncMock--import pytest--from cloud import agent_functions as cloud_agent_functions-from cloud.db.repositories.browser_session_infra import FIRST_PARTY, SessionInfra-from cloud.webeye.browser_types import BrowserInfraProvider---async def test_vendor_session_streams_cdp(monkeypatch: pytest.MonkeyPatch) -> None:- monkeypatch.setattr(- cloud_agent_functions.cloud_db.browser_session_infra,- "get_browser_session_infra",- AsyncMock(return_value=SessionInfra(provider=EXTERNAL_PROVIDER)),- )- fn = cloud_agent_functions.CloudAgentFunction()- assert await fn.resolve_stream_transport(browser_session_id="pbs_1", organization_id="org_1") == "cdp"---async def test_first_party_session_uses_deployment_setting(monkeypatch: pytest.MonkeyPatch) -> None:- monkeypatch.setattr(- cloud_agent_functions.cloud_db.browser_session_infra,- "get_browser_session_infra",- AsyncMock(return_value=FIRST_PARTY),- )- monkeypatch.setattr(cloud_agent_functions.settings, "BROWSER_STREAMING_MODE", "vnc")- fn = cloud_agent_functions.CloudAgentFunction()- assert await fn.resolve_stream_transport(browser_session_id="pbs_1", organization_id="org_1") == "vnc"---async def test_infra_lookup_failure_falls_back_to_setting(monkeypatch: pytest.MonkeyPatch) -> None:- monkeypatch.setattr(- cloud_agent_functions.cloud_db.browser_session_infra,- "get_browser_session_infra",- AsyncMock(side_effect=RuntimeError("db down")),- )- monkeypatch.setattr(cloud_agent_functions.settings, "BROWSER_STREAMING_MODE", "vnc")- fn = cloud_agent_functions.CloudAgentFunction()- assert await fn.resolve_stream_transport(browser_session_id="pbs_1", organization_id="org_1") == "vnc"---async def test_no_session_id_uses_deployment_setting(monkeypatch: pytest.MonkeyPatch) -> None:- monkeypatch.setattr(cloud_agent_functions.settings, "BROWSER_STREAMING_MODE", "vnc")- fn = cloud_agent_functions.CloudAgentFunction()- assert await fn.resolve_stream_transport(browser_session_id=None, organization_id="org_1") == "vnc"-```--Adjust the class name if the override class in `cloud/agent_functions.py` is not `CloudAgentFunction` — use whatever class holds `supports_live_view`, and construct it the way `test_browser_session_connect_url.py` does.--- [ ] **Step 2: Run tests to verify they fail**--Run: `uv run pytest tests/cloud/test_stream_transport_resolution.py -v`-Expected: FAIL — the override does not exist, so the base returns the setting and the vendor test asserts `"cdp"` ≠ `"vnc"`.--- [ ] **Step 3: Implement the override**--In `cloud/agent_functions.py`, directly below `supports_live_view`:--```python- async def resolve_stream_transport(- self, *, browser_session_id: str | None, organization_id: str | None- ) -> str:- """A vendor-held browser has no websockify sidecar to relay RFB from, but its CDP endpoint- is reachable through the session router — so it streams via CDP screencast. First-party- sessions keep the deployment default."""- if browser_session_id:- try:- infra = await cloud_db.browser_session_infra.get_browser_session_infra(browser_session_id)- except Exception:- LOG.warning(- "Could not resolve session infrastructure for stream transport; using the deployment default",- browser_session_id=browser_session_id,- exc_info=True,- )- return settings.BROWSER_STREAMING_MODE- if infra.is_vendor:- return "cdp"- return settings.BROWSER_STREAMING_MODE-```--- [ ] **Step 4: Run the cloud suites**--Run: `uv run pytest tests/cloud/test_stream_transport_resolution.py tests/cloud/test_vendor_identity_nondisclosure.py tests/cloud/test_browser_session_infra.py -v`-Expected: PASS.--- [ ] **Step 5: Compile check and commit**--```bash-uv run python -m py_compile cloud/agent_functions.py-git add cloud/agent_functions.py tests/cloud/test_stream_transport_resolution.py-git commit -m "feat(SKY-13291): stream vendor-held sessions via CDP screencast transport"-```-------### Task 6: Frontend — session type, transport resolver, and hook--**Files:**-- Modify: `skyvern-frontend/src/routes/workflows/types/browserSessionTypes.ts` (line ~18, next to `vnc_streaming_supported`)-- Modify: `skyvern-frontend/src/hooks/useRuntimeConfig.ts`-- Test: `skyvern-frontend/src/hooks/useRuntimeConfig.test.ts` (extend)--**Interfaces:**-- Consumes: `BrowserSessionResponse.stream_transport` from Task 4; existing `normalizeBrowserStreamingMode`, `useBrowserStreamingMode`, `getClient`, `useCredentialGetter`.-- Produces (Task 7 relies on these exact names):- - `resolveStreamTransport(globalMode: BrowserStreamingMode, sessionTransport: string | null | undefined): BrowserStreamingMode` (pure, exported)- - `useStreamTransport(browserSessionId?: string | null): { streamTransport: BrowserStreamingMode }` (exported hook; falls back to the global mode while loading or when the id is absent)- - `BrowserSessionType.stream_transport?: string | null`--- [ ] **Step 1: Write the failing test**--Append to `skyvern-frontend/src/hooks/useRuntimeConfig.test.ts`:--```ts-import { resolveStreamTransport } from "./useRuntimeConfig";--describe("resolveStreamTransport", () => {- it("prefers the session's transport over the global mode", () => {- expect(resolveStreamTransport("vnc", "cdp")).toBe("cdp");- expect(resolveStreamTransport("cdp", "vnc")).toBe("vnc");- });-- it("falls back to the global mode when the session gives none or garbage", () => {- expect(resolveStreamTransport("vnc", null)).toBe("vnc");- expect(resolveStreamTransport("vnc", undefined)).toBe("vnc");- expect(resolveStreamTransport("cdp", "")).toBe("cdp");- expect(resolveStreamTransport("vnc", "webrtc")).toBe("vnc");- });-});-```--- [ ] **Step 2: Run test to verify it fails**--Run: `cd skyvern-frontend && npx vitest run src/hooks/useRuntimeConfig.test.ts`-Expected: FAIL — `resolveStreamTransport` is not exported.--- [ ] **Step 3: Implement type, resolver, and hook**--`browserSessionTypes.ts` — add below `vnc_streaming_supported`:--```ts- stream_transport?: string | null;-```--`useRuntimeConfig.ts` — add (and export) below `useBrowserStreamingMode`; add `import { useCredentialGetter } from "@/hooks/useCredentialGetter";` (same import path `BrowserSession.tsx` uses):--```ts-function resolveStreamTransport(- globalMode: BrowserStreamingMode,- sessionTransport: string | null | undefined,-): BrowserStreamingMode {- if (!sessionTransport || !STREAMING_MODES.has(sessionTransport.trim().toLowerCase())) {- return globalMode;- }- return normalizeBrowserStreamingMode(sessionTransport);-}--function useStreamTransport(browserSessionId?: string | null) {- const { browserStreamingMode } = useBrowserStreamingMode();- const credentialGetter = useCredentialGetter();- const query = useQuery<{ stream_transport?: string | null }>({- queryKey: ["browserSession", browserSessionId],- queryFn: async () => {- const client = await getClient(credentialGetter, "sans-api-v1");- return client- .get(`/browser_sessions/${browserSessionId}`)- .then((response) => response.data);- },- enabled: Boolean(browserSessionId),- staleTime: 5 * 60 * 1000,- refetchOnWindowFocus: false,- retry: 1,- });-- return {- streamTransport: resolveStreamTransport(- browserStreamingMode,- query.data?.stream_transport,- ),- };-}-```--Add both to the file's export list. Note: the `["browserSession", id]` queryKey deliberately matches `BrowserSession.tsx`'s existing session query so react-query dedupes the fetch on pages that already load the session.--- [ ] **Step 4: Run tests and type check**--Run: `cd skyvern-frontend && npx vitest run src/hooks/useRuntimeConfig.test.ts && npx tsc --noEmit`-Expected: PASS / clean.--- [ ] **Step 5: Commit**--```bash-git add skyvern-frontend/src/hooks/useRuntimeConfig.ts skyvern-frontend/src/hooks/useRuntimeConfig.test.ts \- skyvern-frontend/src/routes/workflows/types/browserSessionTypes.ts-git commit -m "feat(SKY-13291): per-session stream transport resolver and hook"-```-------### Task 7: Frontend — switch the stream-selection call sites--**Files (all Modify):**-- `skyvern-frontend/src/routes/browserSessions/BrowserSession.tsx` (~line 51)-- `skyvern-frontend/src/routes/workflows/studio/StreamPresenter.tsx` (~line 30)-- `skyvern-frontend/src/routes/workflows/studio/runview/RunLiveStream.tsx` (~line 26)-- `skyvern-frontend/src/routes/tasks/detail/TaskActions.tsx` (~line 130)-- `skyvern-frontend/src/routes/workflows/workflowRun/WorkflowRunOverview.tsx` (~lines 51, 123)-- `skyvern-frontend/src/routes/workflows/editor/Workspace.tsx` (~lines 532, 582–585)--**Interfaces:**-- Consumes: `useStreamTransport(browserSessionId)` and `resolveStreamTransport(...)` from Task 6.-- Produces: no new symbols. `BrowserPaneHeader.tsx`'s `StreamModeBadge` intentionally keeps the global mode — it is a DEV-only diagnostic of deployment config, not a per-session control.--Per-site edits (each is: swap the hook, keep the rest of the expression identical):--- [ ] **Step 1: BrowserSession.tsx** — it already fetches the full session object, so use the pure resolver, not the fetching hook. Replace lines 51–52:--```tsx- const { browserStreamingMode } = useBrowserStreamingMode();- const isCdpMode = browserStreamingMode === "cdp";-```--with (the `browserSession` query result already exists below; move this derivation after it):--```tsx- const { browserStreamingMode } = useBrowserStreamingMode();- const isCdpMode =- resolveStreamTransport(browserStreamingMode, browserSession?.stream_transport) === "cdp";-```--- [ ] **Step 2: StreamPresenter.tsx** — replace lines 30–31:--```tsx- const { browserStreamingMode } = useBrowserStreamingMode();- const useCdp = browserStreamingMode === "cdp" && !isRecording;-```--with:--```tsx- const { streamTransport } = useStreamTransport(browserSessionId);- const useCdp = streamTransport === "cdp" && !isRecording;-```--- [ ] **Step 3: RunLiveStream.tsx** — replace lines 26 and 34:--```tsx- const { streamTransport } = useStreamTransport(browserSessionId);- ...- const useVnc =- Boolean(browserSessionId) && streamTransport !== "cdp" && !vncFailed;-```--- [ ] **Step 4: TaskActions.tsx** — replace lines 130–131 (note `browserSessionId` is declared on line 129, before the current hook call):--```tsx- const { streamTransport } = useStreamTransport(browserSessionId);- const shouldUseCdpStream = streamTransport === "cdp";-```--- [ ] **Step 5: WorkflowRunOverview.tsx** — line 62 already has `const browserSessionId = workflowRun?.browser_session_id;`. Replace line 51's hook and line 123:--```tsx- const { streamTransport } = useStreamTransport(browserSessionId);- ...- const shouldUseCdpStream = streamTransport === "cdp";-```--Move the hook call below the `workflowRun` query so `browserSessionId` is in scope (hooks order stays unconditional).--- [ ] **Step 6: Workspace.tsx** — the active debug session provides the id. Replace line 532's hook and lines 581–585:--```tsx- const { streamTransport } = useStreamTransport(activeDebugSession?.browser_session_id);- ...- const isCdpStreamingMode = streamTransport === "cdp" && !recordingStore.isRecording;- // Record Browser exfiltration requires VNC even when the transport is CDP streaming.- const preferVncStream = streamTransport !== "cdp" || recordingStore.isRecording;-```--If `activeDebugSession` is declared after line 532, move the hook call to just after that declaration (unconditional call order preserved). Update the comment at line ~2547 (`mode comes from BROWSER_STREAMING_MODE / runtime config`) to `mode comes from the session's stream transport, falling back to runtime config`.--- [ ] **Step 7: Type check, lint, and grep for leftovers**--```bash-cd skyvern-frontend && npx tsc --noEmit && npx eslint \- src/routes/browserSessions/BrowserSession.tsx src/routes/workflows/studio/StreamPresenter.tsx \- src/routes/workflows/studio/runview/RunLiveStream.tsx src/routes/tasks/detail/TaskActions.tsx \- src/routes/workflows/workflowRun/WorkflowRunOverview.tsx src/routes/workflows/editor/Workspace.tsx-grep -rn "useBrowserStreamingMode" src/routes | grep -v BrowserPaneHeader-```--Expected: clean; the grep returns only `BrowserSession.tsx` (resolver pattern keeps the global hook there) and nothing else outside the intentionally-global badge.--- [ ] **Step 8: Commit**--```bash-git add skyvern-frontend/src/routes/browserSessions/BrowserSession.tsx \- skyvern-frontend/src/routes/workflows/studio/StreamPresenter.tsx \- skyvern-frontend/src/routes/workflows/studio/runview/RunLiveStream.tsx \- skyvern-frontend/src/routes/tasks/detail/TaskActions.tsx \- skyvern-frontend/src/routes/workflows/workflowRun/WorkflowRunOverview.tsx \- skyvern-frontend/src/routes/workflows/editor/Workspace.tsx-git commit -m "feat(SKY-13291): select live-stream component per session transport"-```-------### Task 8: Capability flip — vendors serve live view (rollout switch, own PR)--**Files:**-- Modify: `cloud/webeye/browser_types.py` (`PROVIDER_CAPABILITIES`, ~lines 133–145, and the `live_view` comment ~line 121)-- Test: `tests/cloud/test_pbs_infra_routing.py` (extend/adjust)--**Interfaces:**-- Consumes: everything above — this is the switch that routes `needs_live_view=True` sessions to vendors AND (through the existing cloud `supports_live_view`, which returns `capabilities_for(infra.provider).live_view`) makes `vnc_streaming_supported=True` on vendor session responses.-- Produces: `live_view is True` in `PROVIDER_CAPABILITIES` for every external provider.--**Do not merge until Tasks 1–7 are deployed and a staging vendor session shows a working stream.** Reverting this one PR restores the pre-feature routing exactly.--- [ ] **Step 1: Write/adjust the failing tests**--In `tests/cloud/test_pbs_infra_routing.py`, find the test asserting `needs_live_view` forces first-party (capability override) and split the assertion:--```python-async def test_needs_live_view_no_longer_forces_first_party() -> None:- for provider in EXTERNAL_PROVIDERS:- assert capabilities_for(provider).live_view is True-```--Also update (do not delete) any existing test that pinned `live_view=False` — its name documents the old deliberate design; rewrite it to pin the new one.--- [ ] **Step 2: Run tests to verify they fail**--Run: `uv run pytest tests/cloud/test_pbs_infra_routing.py -v`-Expected: FAIL on the new assertion.--- [ ] **Step 3: Flip the table**--In `PROVIDER_CAPABILITIES`, set `live_view=True` for every external provider entry. Replace the `live_view` field comment (lines ~118–121) with:--```python- # First-party sessions stream RFB from the pod's websockify sidecar; vendor-held sessions- # stream CDP screencast through the session router (resolve_stream_transport decides per row).- live_view: bool-```--- [ ] **Step 4: Run the cloud suites**--Run: `uv run pytest tests/cloud/test_pbs_infra_routing.py tests/cloud/test_stream_transport_resolution.py tests/cloud/test_pbs_vendor_session_lifecycle.py -v`-Expected: PASS.--- [ ] **Step 5: Compile check and commit**--```bash-uv run python -m py_compile cloud/webeye/browser_types.py-git add cloud/webeye/browser_types.py tests/cloud/test_pbs_infra_routing.py-git commit -m "feat(SKY-13291): vendors serve live view via CDP screencast transport"-```-------## End-to-end verification (before the Task 8 PR merges)--1. Backend: `uv run pytest tests/unit/forge/sdk/routes/streaming/ tests/unit_tests/test_streaming_screencast.py tests/unit/webeye/test_browser_session_response.py tests/cloud/ -q` and `uv run alembic check` (no schema change expected).-2. Local smoke (OSS mode unchanged): `BROWSER_STREAMING_MODE=cdp ./run_skyvern.sh`, open a browser session page, confirm the screencast still streams (base hook returns the global setting — behavior identical to today).-3. Staging smoke (the new path): create a vendor-routed session via the API (`DYNAMIC_BROWSER_TYPE` flag arm or the infra flag), GET `/browser_sessions/{id}` → `stream_transport == "cdp"` and `vnc_streaming_supported == true` (after Task 8), open the session page → `InteractiveStreamView` renders frames. Capture before/after screenshots for the FE PR.-4. Watch router load during the staging soak: screencast frames traverse the cdp-proxy (`skyvern.cdp_proxy.*` metrics); SKY-12502 (frame throttle/downscale via the policy engine) is the companion if per-viewer byte rates need bounding.--## Known risks--- `useStreamTransport` adds one `/browser_sessions/{id}` fetch on run pages that didn't previously load the session (deduped elsewhere via the shared queryKey). Acceptable; revisit only if endpoint latency shows up.-- The vendor's own operator UI could theoretically hold a screencast on the same page target; the spike saw no interference, but if frames stall while automation runs, check for a competing `Page.startScreencast` client before blaming the router.-- One provider's `cdp_url` is `http(s)` requiring `/json/version` discovery — irrelevant here (Playwright's `connect_over_cdp` handles it, and the API server dials the routed `browser_address`), but any future raw-websocket consumer must replicate it.diff --git a/docs/plans/sky-13117-run-count-axis-scale.md b/docs/plans/sky-13117-run-count-axis-scale.mddeleted file mode 100644index 10609186a..000000000--- a/docs/plans/sky-13117-run-count-axis-scale.md+++ /dev/null@@ -1,21 +0,0 @@-# SKY-13117: Keep run-count axis units consistent--## Scope--- Update `skyvern-frontend/cloud/routes/analytics/components/RunStatusHeroChart.tsx`- so every run-count tick on a rendered axis uses one shared magnitude suffix.-- Update- `skyvern-frontend/cloud/routes/analytics/components/RunStatusHeroChart.test.tsx`- with regression coverage for a thousands-scale axis ending at one million.--## Acceptance criteria--- A thousands-scale axis renders `250K`, `500K`, `750K`, and `1,000K`.-- A single axis never switches from `K` to `M` between ticks.-- Small run-count axes remain locale-formatted without a suffix.-- The count axis remains wide enough for the longest expected compact label.--## Verification--- Run the focused `RunStatusHeroChart` Vitest test.-- Run ESLint on the two changed frontend source files.diff --git a/docs/screenshots/schedule-button-placement-before.png b/docs/screenshots/schedule-button-placement-before.pngdeleted file mode 100644index caefd1841..000000000Binary files a/docs/screenshots/schedule-button-placement-before.png and /dev/null differdiff --git a/docs/screenshots/schedule-button-placement.png b/docs/screenshots/schedule-button-placement.pngdeleted file mode 100644index e2609e94b..000000000Binary files a/docs/screenshots/schedule-button-placement.png and /dev/null differdiff --git a/docs/screenshots/sky-10794-2fa-login-block/00-before-always-shown.png b/docs/screenshots/sky-10794-2fa-login-block/00-before-always-shown.pngdeleted file mode 100644index 8298cd597..000000000Binary files a/docs/screenshots/sky-10794-2fa-login-block/00-before-always-shown.png and /dev/null differdiff --git a/docs/screenshots/sky-10794-2fa-login-block/01-after-collapsed.png b/docs/screenshots/sky-10794-2fa-login-block/01-after-collapsed.pngdeleted file mode 100644index 511358f00..000000000Binary files a/docs/screenshots/sky-10794-2fa-login-block/01-after-collapsed.png and /dev/null differdiff --git a/docs/screenshots/sky-10794-2fa-login-block/02-after-expanded.png b/docs/screenshots/sky-10794-2fa-login-block/02-after-expanded.pngdeleted file mode 100644index 7553cea46..000000000Binary files a/docs/screenshots/sky-10794-2fa-login-block/02-after-expanded.png and /dev/null differdiff --git a/docs/screenshots/sky-10794-2fa-login-block/03-after-saved-values.png b/docs/screenshots/sky-10794-2fa-login-block/03-after-saved-values.pngdeleted file mode 100644index c908dafe5..000000000Binary files a/docs/screenshots/sky-10794-2fa-login-block/03-after-saved-values.png and /dev/null differdiff --git a/docs/screenshots/sky-10794-2fa-login-block/04-after-credential-summary.png b/docs/screenshots/sky-10794-2fa-login-block/04-after-credential-summary.pngdeleted file mode 100644index a29e977b1..000000000Binary files a/docs/screenshots/sky-10794-2fa-login-block/04-after-credential-summary.png and /dev/null differ