@mindrian_os/cli 1.15.3-beta.44 → 1.15.3-beta.48

package/CHANGELOG.md

@@ -1,3 +1,112 @@
+## [1.15.3-beta.48] - 2026-07-26
+
+### Added
+- **Room-graph density read (Phase 232.1).** `/mos:doctor` and `room_state status`/`get-state`
+  now self-report node/edge counts per `room.db`, read exclusively through a new read-only
+  `navigation.cjs` door (`openRoomDbReadOnlyForCaller`) -- never the mutating door, never a
+  direct `room.db` open. Closes SEED-074's own "suggested first move"
+  (`.planning/seeds/SEED-074-local-graph-read-layer-lacks-salience-and-query-time-joins.md`):
+  the seed's actual PageRank/Louvain/query-time-join target stays gated exactly as written;
+  this only makes its own trigger condition (measured room.db density) self-reported instead
+  of something a human has to remember to check by hand. No output string anywhere claims a
+  room graph is dense, at-risk, or healthy (hard guard, grep-verified). Goal-backward
+  verification PASS, 6/6 must-haves (`232.1-VERIFICATION.md`).
+- Local reified-claim `ContradictionEvent` primitive (quick task 260725-9ca), with a hermetic
+  acceptance test.
+
+### Fixed
+- **`room_bind` could report `ok:true` while a sibling MCP read tool silently resolved a
+  stale, unrelated room.** Root cause: 20 of 21 MCP tool call sites were missing the
+  `CLAUDE_CODE_SESSION_ID` stdio fallback that `room_bind` itself already had, so a
+  session-id mismatch between the bind call and a later read call could bind one room but
+  read another without ever surfacing an error. Fixed via one shared
+  `resolveEffectiveSessionId` helper, wired into all 21 sites; live-reproduced, then closed,
+  tests green. Full reproduction and root cause also filed in the `rethinking-mindrianos`
+  room per this repo's dev-research compositing rule.
+- **Phase 232.1's own room-graph density census could silently drop a room that used the
+  `abs_path` registry field instead of `path`,** undermining the one thing the census exists
+  to be accurate about. `resolveRoomPath` now checks `abs_path` first, `path` as fallback,
+  matching the precedence every other production call site in this codebase already honors;
+  pinned by a mutation-verified regression test (reverted, confirmed red, restored, confirmed
+  green). The identical gap in the earlier, already-shipped `cascade-rooms-module.cjs` this
+  was copied from is left alone -- out of this phase's scope, tracked separately as
+  low-severity follow-up debt.
+
+## [1.15.3-beta.46] - 2026-07-23
+
+### Added
+- 
+
+### Fixed
+- **The F.8 "bind session to room" Decision Gate (and any PRIMARY-path registry gate)
+  force-fired on turns whose ONLY content was an automated background-task-completion
+  notification, with zero real user text, blocking continuation with a Stop hook error even
+  though there was nothing for the navigator to decide that turn.** Root cause:
+  `precedingUserText` resolves to `''` for two very different reasons that
+  `scripts/check-card-fire.cjs` could not previously distinguish -- a genuinely terse HUMAN
+  turn ("ok", "go on") and a SYNTHETIC preceding transcript record (a `tool_result` envelope
+  from a background tool call, or an automated task-notification block) with no
+  human-authored text at all. `lib/core/gate-relevance.cjs`'s conservative low-signal branch
+  forced (assumed relevant) on both, which is only correct for the first -- there is no human
+  turn for a Decision Gate to be relevant or irrelevant to on the second. Fixed by adding a
+  new `preceding_user_text_source` signal (`'typed' | 'tool_result' | 'none'`), classified by
+  a new `classifyPrecedingUserContentSource` helper in `readTranscriptTurn` and threaded
+  through `deriveTurnSignals`; `classifyCardFire`'s PRIMARY-path relevance branch now bypasses
+  forcing immediately when the source is confirmed `'tool_result'`
+  (`reason: 'preceding-turn-synthetic-no-user-engagement'`), before ever reaching the
+  conservative low-signal branch. A genuinely terse human turn against the identical gate
+  still force-fires, unweakened (the WR-06/CR-06 floor). Live-reproduced 3 consecutive times
+  in one session; the mechanism itself was first diagnosed 2026-07-06 (as a fact, not yet
+  treated as a defect) and confirmed recurring 2026-07-11 and 2026-07-22. Full RCA:
+  `.planning/debug/resolved/room-bind-gate-fires-on-notification-only-turns.md`.
+- **The semantic-edge derivation queue (`graph-derive-queue.json`) silently cleared itself
+  on every failed derivation, in every room, forever -- so no room has ever gotten a real
+  INFORMS/CONTRADICTS/CONVERGES cascade edge.** Root cause: `scripts/gsd-graph-derive-drain.cjs`'s
+  `drainDerive()` caught a `runDerivation` throw, pushed the failed room to `drained` anyway,
+  and rewrote the queue empty with no log -- destroying the retry signal on every SessionStart
+  run regardless of outcome. Two of the three originally-stacked causes (a dead standalone
+  Anthropic API key; the headless path computing a null single-pair derive with no real
+  artifact pairs) had already been resolved upstream by Phase 224-02's switch to local
+  embeddings; the silent-clear was the one still live. Fixed via a new `reconcileQueue()`: a
+  failed room is now kept and retried (capped at `MAX_DERIVE_ATTEMPTS=5`, then dropped
+  `permanent:true` and logged, never silently), and every failure appends to a new
+  `<room>/.mindrian/graph-derive-failures.json`. Root-caused by cross-referencing a TDS
+  "context rot" research thread against the live debug queue, not filed fresh. Full RCA:
+  `.planning/debug/graph-derive-silent-clear-dead-api-derivation.md`. Formalizing the
+  derivation transport, healing the ~16 already-damaged rooms, and a doctor health check
+  remain open (register items 4b-4e).
+- **`room_search` applied its 50-result cap in raw filesystem-walk order, before any
+  relevance ranking -- so a genuinely relevant match in a late-traversed folder could lose to
+  50 incidental early-folder hits, and a same-entity query could return dozens of
+  near-duplicate lines from one file.** Root cause: `lib/mcp/tools/room.cjs`'s `searchRoom`
+  was a plain `String.includes` grep that pushed matches in directory-entry order and
+  returned the instant either `SEARCH_MAX_RESULTS` (50) or `SEARCH_MAX_FILES` (500) was hit --
+  capped-then-never-ranked, not ranked-then-capped. Fixed via a rank-then-cap rewrite
+  (`collectMatches`/`rankMatches`): match density x0.7 + recency x0.3, with a 5-per-file slice
+  cap so one file can't monopolize the result budget. `graph_query` was audited in the same
+  pass and found already ranked correctly (a composite relevance score already lives in
+  `navigation/neighborhood.cjs`); `whitespace_scan` is unranked but uncapped, deferred pending
+  the graph-derive edge-density fix above (little signal to rank while most rooms carry
+  near-zero semantic edges). Full RCA: `.planning/debug/graph-query-results-unranked.md`.
+- **The M:OS Canonical Design System v1.1 "bake into all HTML artifacts" mandate was only
+  ~30% actually landed, despite its own commit (`a9e1ee88`) and Phase 232-01 claiming it was
+  done.** The CSS bundle, loader (`mosStyleTag()`), and `lib/wiki/wiki-layout.cjs`'s
+  retokenization were real; `scripts/generate-deck.cjs`, `generate-hub.cjs`,
+  `generate-lobby.cjs`, and `generate-snapshot.cjs` had zero reference to `mosStyleTag()`,
+  `dashboard/index.html` had zero M:OS tokens, and the mandate's own doc
+  (`skills/ui-system/rules/design-system.md`) and `SKILL.md` section 0 did not exist. Wired
+  all 4 generators to `mosStyleTag()` (cream default, `data-theme="light"`), injected canonical
+  tokens plus a role-based CSS variable alias layer into `dashboard/index.html` and
+  `dashboard/export-template.html` (legacy `--mondrian-*`/`--ds-*` names aliased onto canonical
+  values in place, not renamed), and authored the missing mandate docs. Code review caught 2
+  real regressions before this shipped: (1) 3 of the 4 generators' own pre-existing `<style>`
+  blocks redeclared the same token names with old hex values LATER in the document, so the
+  mandate rendered in the markup but had zero visual effect by CSS cascade -- fixed by removing
+  the colliding redeclarations; (2) `dashboard/index.html`'s dark-to-light polarity flip broke
+  hardcoded Cytoscape graph-label colors and hover overlays tuned for the old dark theme --
+  fixed by repointing them at the resolved ink values. Both independently re-verified via
+  Playwright (`getComputedStyle`) against live-regenerated output, not just diffs.
+
 ## [1.15.3-beta.44] - 2026-07-23
 
 ### Added

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@mindrian_os/cli",
-  "version": "1.15.3-beta.44",
+  "version": "1.15.3-beta.48",
   "description": "Install MindrianOS into Claude Code with one command -- `npx @mindrian_os/cli`. Ships the MindrianOS plugin (Larry + PWS methodology + Data Room) plus a setup/diagnostics CLI (install/doctor/update).",
   "scripts": {
     "mcp": "node bin/mindrian-mcp-server.cjs",