Skip to content

Garry's Opinionated OpenClaw/Hermes Agent Brain

TypeScript

v0.60.667.0: stamp

garrytan2aa0ca387ec7
diff --git a/.claude-plugin/plugin.json b/.claude-plugin/plugin.json
index 478ef296a5..f072824363 100644
--- a/.claude-plugin/plugin.json
+++ b/.claude-plugin/plugin.json
@@ -1,6 +1,6 @@
{
"name": "gbrain",
- "version": "0.60.666.0",
+ "version": "0.60.667.0",
"description": "Personal knowledge brain for your coding agent — hybrid search, synthesis, graph traversal, and durable cross-session memory over Postgres/PGLite with pgvector, plus a curated brain-first skill set.",
"author": {
"name": "Garry Tan",
diff --git a/.codex-plugin/plugin.json b/.codex-plugin/plugin.json
index 0a97d396b7..c955952126 100644
--- a/.codex-plugin/plugin.json
+++ b/.codex-plugin/plugin.json
@@ -1,6 +1,6 @@
{
"name": "gbrain",
- "version": "0.60.666.0",
+ "version": "0.60.667.0",
"description": "Personal knowledge brain for your coding agent — hybrid search, synthesis, graph traversal, and durable cross-session memory over Postgres/PGLite with pgvector, plus a curated brain-first skill set.",
"author": {
"name": "Garry Tan",
diff --git a/BOOTSTRAP_FOR_AGENTS.md b/BOOTSTRAP_FOR_AGENTS.md
index 7b0de83522..a5a0ba5388 100644
--- a/BOOTSTRAP_FOR_AGENTS.md
+++ b/BOOTSTRAP_FOR_AGENTS.md
@@ -1,4 +1,4 @@
-<!-- gbrain-runbook-stamp: 0.60.666.0 -->
+<!-- gbrain-runbook-stamp: 0.60.667.0 -->
<!-- This stamp must equal the VERSION file at every release; CI enforces it
(scripts/check-bootstrap-tag.sh). `gbrain bootstrap status` compares it to
the installed binary and warns on skew. -->
diff --git a/CHANGELOG.md b/CHANGELOG.md
index 4552a2acf6..67c45ca46c 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -10,7 +10,7 @@ credits are retained; no result has been reassigned to another provider. Origina
identifiers and attribution are available in the pre-removal Git revision
`6040075c6cb95be5881cc2e1b76ef7d71f4e5d29` (retained on 2026-09-23).
-## [Unreleased]
+## [0.60.667.0] - 2026-11-11
**A brain that serves several people can now be set up the safe way in one command: each person gets a member, a folder only they read and a key bound to them, and doctor fails when a person's folder has no owner.** When one bot answers several people from one brain with one unbound key, it pulls one person's notes into another person's answers (garrytan/harness-evals#363 measured 15 to 47.5% of trials; bots reading through member-bound keys stayed at 0 to 0.5%). `gbrain setup claude-code --serves several --members alice-example,bob-example` writes `access.serves = several`, adds the members, adds a folder rule `people/<member>/` per member through the folder-rule path (committed to `.gbrain/access.yml` on a repo brain, the database rule log on a repo-less one), mints one read-write key per member through the `auth create` mint (granted the labelled source) into a 0600 file under `~/.gbrain/member-keys/`, and proves by reading every other member's folder as each member that none is visible. Re-runs keep everything and never replace a key; a partial failure finishes on the next run, and anything the preview marks BLOCKED stops the run before it writes. A one-person brain sees no change: `--serves one` stays the default and its output is unchanged.
diff --git a/VERSION b/VERSION
index 02597673d0..73195b2d0c 100644
--- a/VERSION
+++ b/VERSION
@@ -1 +1 @@
-0.60.666.0
+0.60.667.0
diff --git a/openclaw.plugin.json b/openclaw.plugin.json
index 7822b003af..a242e336bc 100644
--- a/openclaw.plugin.json
+++ b/openclaw.plugin.json
@@ -1,7 +1,7 @@
{
"id": "gbrain-context-engine",
"name": "gbrain",
- "version": "0.60.666.0",
+ "version": "0.60.667.0",
"description": "Personal knowledge brain with Postgres + pgvector hybrid search",
"family": "bundle-plugin",
"kind": "context-engine",
diff --git a/package.json b/package.json
index fd7b95bb37..ce5c3af965 100644
--- a/package.json
+++ b/package.json
@@ -254,7 +254,7 @@
"bun": ">=1.4.4"
},
"license": "MIT",
- "version": "0.60.666.0",
+ "version": "0.60.667.0",
"overrides": {
"@aws-sdk/util-format-url": "3.972.9",
"@ai-sdk/provider-utils": "4.0.56",
diff --git a/plugin-variants/gbrain-coding/.claude-plugin/plugin.json b/plugin-variants/gbrain-coding/.claude-plugin/plugin.json
index 49fd030e3d..62b027d30f 100644
--- a/plugin-variants/gbrain-coding/.claude-plugin/plugin.json
+++ b/plugin-variants/gbrain-coding/.claude-plugin/plugin.json
@@ -1,6 +1,6 @@
{
"name": "gbrain-coding",
- "version": "0.60.666.0",
+ "version": "0.60.667.0",
"description": "Brain-first coding agent working inside a repo: retrieval, routing, ingest discipline, correction hygiene. Default persona for the claude-code harness bridge; also published as the gbrain-coding marketplace variant. (persona variant of the gbrain plugin — 21 skills)",
"author": {
"name": "Garry Tan",
diff --git a/plugin-variants/gbrain-coding/.codex-plugin/plugin.json b/plugin-variants/gbrain-coding/.codex-plugin/plugin.json
index f84b887b12..2216f7a55e 100644
--- a/plugin-variants/gbrain-coding/.codex-plugin/plugin.json
+++ b/plugin-variants/gbrain-coding/.codex-plugin/plugin.json
@@ -1,6 +1,6 @@
{
"name": "gbrain-coding",
- "version": "0.60.666.0",
+ "version": "0.60.667.0",
"description": "Brain-first coding agent working inside a repo: retrieval, routing, ingest discipline, correction hygiene. Default persona for the claude-code harness bridge; also published as the gbrain-coding marketplace variant. (persona variant of the gbrain plugin — 21 skills)",
"author": {
"name": "Garry Tan",
diff --git a/plugin-variants/gbrain-coding/README.md b/plugin-variants/gbrain-coding/README.md
index 2d1e1066d4..4087fc4e77 100644
--- a/plugin-variants/gbrain-coding/README.md
+++ b/plugin-variants/gbrain-coding/README.md
@@ -1,4 +1,4 @@
-<!-- gbrain-plugin-tree-stamp: 0.60.666.0 -->
+<!-- gbrain-plugin-tree-stamp: 0.60.667.0 -->
# gbrain-coding (generated persona variant — do not hand-edit)
Brain-first coding agent working inside a repo: retrieval, routing, ingest discipline, correction hygiene. Default persona for the claude-code harness bridge; also published as the gbrain-coding marketplace variant.
diff --git a/plugin-variants/gbrain-daily/.claude-plugin/plugin.json b/plugin-variants/gbrain-daily/.claude-plugin/plugin.json
index a13f2f40cb..c851324d9e 100644
--- a/plugin-variants/gbrain-daily/.claude-plugin/plugin.json
+++ b/plugin-variants/gbrain-daily/.claude-plugin/plugin.json
@@ -1,6 +1,6 @@
{
"name": "gbrain-daily",
- "version": "0.60.666.0",
+ "version": "0.60.667.0",
"description": "Personal knowledge-brain daily use: meetings, tasks, briefings, reading, research. Published as the gbrain-daily marketplace variant. (persona variant of the gbrain plugin — 20 skills)",
"author": {
"name": "Garry Tan",
diff --git a/plugin-variants/gbrain-daily/.codex-plugin/plugin.json b/plugin-variants/gbrain-daily/.codex-plugin/plugin.json
index 473caa630e..e4bef742d4 100644
--- a/plugin-variants/gbrain-daily/.codex-plugin/plugin.json
+++ b/plugin-variants/gbrain-daily/.codex-plugin/plugin.json
@@ -1,6 +1,6 @@
{
"name": "gbrain-daily",
- "version": "0.60.666.0",
+ "version": "0.60.667.0",
"description": "Personal knowledge-brain daily use: meetings, tasks, briefings, reading, research. Published as the gbrain-daily marketplace variant. (persona variant of the gbrain plugin — 20 skills)",
"author": {
"name": "Garry Tan",
diff --git a/plugin-variants/gbrain-daily/README.md b/plugin-variants/gbrain-daily/README.md
index a7c15d0fba..2f444ae887 100644
--- a/plugin-variants/gbrain-daily/README.md
+++ b/plugin-variants/gbrain-daily/README.md
@@ -1,4 +1,4 @@
-<!-- gbrain-plugin-tree-stamp: 0.60.666.0 -->
+<!-- gbrain-plugin-tree-stamp: 0.60.667.0 -->
# gbrain-daily (generated persona variant — do not hand-edit)
Personal knowledge-brain daily use: meetings, tasks, briefings, reading, research. Published as the gbrain-daily marketplace variant.
diff --git a/plugin/README.md b/plugin/README.md
index 208422d056..941ef2b37f 100644
--- a/plugin/README.md
+++ b/plugin/README.md
@@ -1,4 +1,4 @@
-<!-- gbrain-plugin-tree-stamp: 0.60.666.0 -->
+<!-- gbrain-plugin-tree-stamp: 0.60.667.0 -->
# gbrain plugin skill tree (generated — do not hand-edit)
This tree is the curated skill set for the gbrain Codex and Claude Code
diff --git a/src/core/behavior-change-notice.ts b/src/core/behavior-change-notice.ts
index 3b5dd28b31..67cedf64ee 100644
--- a/src/core/behavior-change-notice.ts
+++ b/src/core/behavior-change-notice.ts
@@ -1521,8 +1521,8 @@ export const BEHAVIOR_CHANGES: ReadonlyArray<{ since: string; text: ChangeText;
{ since: '0.60.666.0', id: 'qm_recall_own_page_contract_v4_gst2407', text: 'Each `memory_recall` fact in `structuredContent.facts` gains `own_page`: `true` when the fact sits on the person\'s own member page in the connection\'s source, the same ownership check `forget_fact` makes before it withdraws a fact, and `false` when `forget_fact` for that attested person would answer `fact_not_found`. `arm` is unchanged, so an own-page fact older than 24 hours is `ranked` with `own_page: true`, and a scope read answers `own_page: false`. `get_write_request` and `cancel_write_request` on a held proposal now also answer `created_at` and `updated_at`. QM contract `docs/contracts/qm-memory-provider/v4.json` adds `memory_propose` with its proposal receipt, states and retention, a `scope_read` marking on each personal tool and `own_page`; v1, v2 and v3 are byte-identical.' },
{ since: '0.60.661.0', id: 'recreated_takes_section_keeps_stored_takes_gst2228', text: 'A takes section that gbrain re-creates on a page whose section a rewrite dropped now starts with the takes still stored for that page. `think --take --anchor <slug>` and `gbrain takes extract --from-pages` (also with `--include-covered` on a managed brain) render the stored takes, resolutions included, into the section they create and number their own takes after them, instead of numbering from 1 over a stored take. A take the user removed, or an explicit empty section retired, stays removed. A page whose stored take has a resolution without a quality, or a cell that would not read back unchanged from the section (such as the literal text `<br>`), is refused (`invalid_input`) instead of changing it, and `takes extract --from-pages` skips a page whose takes markers are malformed (`fence_unparsed`) and goes on to the next page.' },
{ since: '0.60.661.0', id: 'links_rows_keep_own_dates_gst2228', text: 'Each row of a page\'s `## Links` section (target, type, source) now keeps its own `valid_from` and `valid_until`. On a page whose links section a rewrite dropped, `add_link` writes the section back with every kept row and its own dates instead of a section holding only the new link; `remove_link` with a `link_source` deletes that row\'s dates for any source, not only `manual`, and leaves other rows\' dates alone. Dates stored before rows kept their source have none: with several rows for one target and type and the section gone, `add_link` and `remove_link` refuse `link_dates_unattributed` and ask for the section back through `edit_page`, one row per source with its own dates. `append_page` text holding a links fence marker is refused with `append_links_fence`; write each link with `add_link`.' },
- { since: 'TBD', id: 'serves_several_labels_gst363', text: 'A brain now says whether it serves one person or several: config `access.serves` (`one` or `several`), else two or more enabled members means several (`gbrain access serves` shows which and why). `gbrain setup <harness> --serves several --members <id,...>` (or `--members-file`) labels each person: a member, a folder rule `people/<member>/` only that member reads, and one key bound to that member in a 0600 file under `~/.gbrain/member-keys/` (the receipt prints paths, never keys), then proves it with one bound read per member; `gbrain access serves several` does the labelling alone. On such a brain a plain `gbrain setup claude-code|codex` now refuses `setup_serves_several` and writes nothing, unless the harness entry carries `--stdio-member` or config `mcp.stdio_member`, set by hand, binds the connection; setup never writes that brain-wide key, and `--stdio-member <member>` (claude-code or codex, refused where OpenClaw or Hermes is installed) binds only that harness entry to that member; its brain-first block says the connection reads only what the person asking may read. `gbrain doctor` fails the new `unlabeled_people_folders` check when such a brain has a `people/<name>/` folder with pages and no folder rule naming its readers. A one-person brain sees no change.' },
- { since: 'TBD', id: 'stdio_member_per_harness_gst363', text: '`gbrain setup claude-code|codex --stdio-member <member>` now writes `serve --stdio-member <member>` into that one harness entry instead of setting config `mcp.stdio_member`, so only that harness on that machine reads as the member, on PGLite and Postgres alike; setup never writes `mcp.stdio_member`, which still binds, when set by hand, every stdio server started without the flag. The flag no longer needs `--serves several`, the member must be enabled, and setup still refuses it where OpenClaw or Hermes is installed or a plugin serves the harness, and `gbrain serve` refuses to start with it where OpenClaw or Hermes is installed. Re-runs and `gbrain post-upgrade` keep an entry\'s member and never add one; `--no-stdio-member` removes it, also on a brain that serves several people. A disabled or unknown member reads as nobody.' },
+ { since: '0.60.667.0', id: 'serves_several_labels_gst363', text: 'A brain now says whether it serves one person or several: config `access.serves` (`one` or `several`), else two or more enabled members means several (`gbrain access serves` shows which and why). `gbrain setup <harness> --serves several --members <id,...>` (or `--members-file`) labels each person: a member, a folder rule `people/<member>/` only that member reads, and one key bound to that member in a 0600 file under `~/.gbrain/member-keys/` (the receipt prints paths, never keys), then proves it with one bound read per member; `gbrain access serves several` does the labelling alone. On such a brain a plain `gbrain setup claude-code|codex` now refuses `setup_serves_several` and writes nothing, unless the harness entry carries `--stdio-member` or config `mcp.stdio_member`, set by hand, binds the connection; setup never writes that brain-wide key, and `--stdio-member <member>` (claude-code or codex, refused where OpenClaw or Hermes is installed) binds only that harness entry to that member; its brain-first block says the connection reads only what the person asking may read. `gbrain doctor` fails the new `unlabeled_people_folders` check when such a brain has a `people/<name>/` folder with pages and no folder rule naming its readers. A one-person brain sees no change.' },
+ { since: '0.60.667.0', id: 'stdio_member_per_harness_gst363', text: '`gbrain setup claude-code|codex --stdio-member <member>` now writes `serve --stdio-member <member>` into that one harness entry instead of setting config `mcp.stdio_member`, so only that harness on that machine reads as the member, on PGLite and Postgres alike; setup never writes `mcp.stdio_member`, which still binds, when set by hand, every stdio server started without the flag. The flag no longer needs `--serves several`, the member must be enabled, and setup still refuses it where OpenClaw or Hermes is installed or a plugin serves the harness, and `gbrain serve` refuses to start with it where OpenClaw or Hermes is installed. Re-runs and `gbrain post-upgrade` keep an entry\'s member and never add one; `--no-stdio-member` removes it, also on a brain that serves several people. A disabled or unknown member reads as nobody.' },
];
/** The newest disclosed change's release: the notice id moves only when a release adds rows. */
diff --git a/templates/bootstrap/template-repo/README.md b/templates/bootstrap/template-repo/README.md
index 76c2ba30f3..67b6bb5d92 100644
--- a/templates/bootstrap/template-repo/README.md
+++ b/templates/bootstrap/template-repo/README.md
@@ -1,6 +1,6 @@
# gbrain agent workspace — template
-<!-- gbrain-template-stamp: 0.60.666.0 -->
+<!-- gbrain-template-stamp: 0.60.667.0 -->
This repository is the **"Use this template"** distribution artifact for a
[gbrain](https://github.com/garrytan/gbrain) personal-agent workspace — the same