@mindrian_os/install 1.13.0-beta.13 → 1.13.0-beta.14

package/.claude-plugin/plugin.json

@@ -1,7 +1,7 @@
 {
   "name": "mos",
   "description": "MindrianOS -- Your AI innovation co-founder. Larry thinks with you through PWS methodology, builds your Data Room as you explore, and chains frameworks intelligently. Install and go.",
-  "version": "1.13.0-beta.13",
+  "version": "1.13.0-beta.14",
   "author": {
     "name": "Jonathan Sagir",
     "url": "https://mindrian.ai"

package/CHANGELOG.md

@@ -1,3 +1,8 @@
+## [1.13.0-beta.14] - 2026-05-14
+
+### Added
+- 
+
 # Changelog
 
 All notable changes to MindrianOS Plugin will be documented in this file.
@@ -43,7 +48,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - **`data/deployment-surfaces.json`** added -- hand-maintained static manifest; 6 surfaces; reuses the `data/` layout convention from Phase 122 but NOT the generator/`--check` pattern (this file isn't derived from anything; nothing to `--check` it against). Schema extension in Plan-03: optional `path_within_file` field on `exact-value` surfaces points at a JSON sub-field (e.g. `statusLine.command` inside `settings.json`); class-J's `exact-value` check extracts via that path before comparing.
 - **`docs/install/BRAIN-SETUP.md`** + **`.env.brain.template`**: state explicitly that auth is `Authorization: Bearer <key>` only (NOT `x-api-key`); surface the `https://mindrianos.vercel.app/brain-access` URL in the no-key fallback (Plan 123-07).
 - **`docs/CANON-PHASE-MAP.md`**: Phase 123 mapped under **Part 6** (dog-fooding the install lifecycle -- one record + one manifest + one command + one release script; the plugin's own install state honors the plugin's canon) and **Part 7** (reuse justification -- ~90% of Phase 123 extends shipped code; net-new files are `data/deployment-surfaces.json`, `lib/core/resolve-brain-key.cjs`, `lib/core/cache-prune.cjs`, the per-class fixtures) (Plan 123-06 Task 4).
-- **`@mindrian_os/cli` -> `@mindrian_os/install` doc/test sweep** -- forward-facing references across `docs/install/PACKAGING-PATHS.md`, `tests/manual/95.6-windows-cold-install-acceptance.md`, `tests/test-release-npm-gate.sh`, `docs/INSTALL-LIFECYCLE-HARNESS.md` (lines 91/104/124), plus older `@mindrian/os` mentions in `docs/autopsies/2026-05-09-gary-laben-install-failure.md`, `docs/UI-UX-CONVERGENCE-2026-05-10/04-REVERSE-SALIENT-INSTALL.md`, `docs/testers/gary-laben/FEEDBACK.md`, `docs/testers/outbox/2026-05-07-gary-laben-welcome.md`. Historical CHANGELOG entries stay as the historical record. After the sweep, `grep -rln "@mindrian_os/cli" docs/install/ commands/ tests/test-*.sh scripts/release.sh` returns nothing (Plan 123-05).
+- **`@mindrian_os/cli` -> `@mindrian_os/install` doc/test sweep** -- forward-facing references across `docs/install/PACKAGING-PATHS.md`, `tests/manual/95.6-windows-cold-install-acceptance.md`, `tests/test-release-npm-gate.sh`, `docs/INSTALL-LIFECYCLE-HARNESS.md` (lines 91/104/124), plus older `@mindrian/os` mentions in `[private case archive]`, `docs/UI-UX-CONVERGENCE-2026-05-10/04-REVERSE-SALIENT-INSTALL.md`, `[private tester archive]`, `docs/testers/outbox/2026-05-07-gary-laben-welcome.md`. Historical CHANGELOG entries stay as the historical record. After the sweep, `grep -rln "@mindrian_os/cli" docs/install/ commands/ tests/test-*.sh scripts/release.sh` returns nothing (Plan 123-05).
 
 ### Notes
 
@@ -93,11 +98,11 @@ The v1.13.0 CAPSTONE release -- headline content is the **Workflow Layer** (Phas
 
 ## [1.13.0-beta.9] - 2026-05-11
 
-Phase 95.6 -- install-cache Windows hardening + skill-loop resilience. Closes case #4 in the install-cache failure family (the 2026-05-08/09 Gary Laben Windows 11 install). Release infrastructure ships as a beta first per the project release policy; this beta is opt-in only.
+Phase 95.6 -- install-cache Windows hardening + skill-loop resilience. Closes case #4 in the install-cache failure family (the 2026-05-08/09 the Wave-2 tester Windows 11 install). Release infrastructure ships as a beta first per the project release policy; this beta is opt-in only.
 
 ### Fixed
 
-- **install.sh skill-loop no longer aborts on a structurally-incomplete skill (D-03).** A skill directory missing its `SKILL.md` now produces a `WARN: skipping skill <name>: no SKILL.md` on stderr and the install continues, instead of hitting `set -euo pipefail` and halting mid-loop. This is the bug that broke Gary Laben's install live on 2026-05-08/09 (`cp: cannot stat '.../skills/mullins-scaffold//SKILL.md'`, exit 1, leaving agents / hooks / settings.json / larry-extended-default unwritten). Also backfilled the missing `skills/mullins-scaffold/SKILL.md` data file.
+- **install.sh skill-loop no longer aborts on a structurally-incomplete skill (D-03).** A skill directory missing its `SKILL.md` now produces a `WARN: skipping skill <name>: no SKILL.md` on stderr and the install continues, instead of hitting `set -euo pipefail` and halting mid-loop. This is the bug that broke the Wave-2 tester's install live on 2026-05-08/09 (`cp: cannot stat '.../skills/mullins-scaffold//SKILL.md'`, exit 1, leaving agents / hooks / settings.json / larry-extended-default unwritten). Also backfilled the missing `skills/mullins-scaffold/SKILL.md` data file.
 - **Windows long-path failures during `git clone` (D-01).** install.sh now detects Windows + Git Bash, runs `git config --global core.longpaths true` before the clone (with an explanatory banner about the global git-config change), cleans a stale partial-clone directory before a retry, and fails with a clear "Git for Windows" error + download link if git is missing on Windows. OS detection is testable on Linux CI via `MOS_TEST_FORCE_OS`.
 - **The "silent install-incomplete" failure mode (D-09).** install.sh now registers the statusLine block FIRST (idempotently, via `register_statusline()`) so a later skill-loop WARN can never again leave the bottom-of-terminal statusline unstamped; it writes an `.install-receipt.json` so a halted install is detectable; `/mos:doctor` gained a class H (`install-incomplete`) that detects a missing statusLine block, reads the receipt, and `--fix` re-stamps the block idempotently; a fresh first session auto-runs `/mos:doctor` and surfaces all-green or names what is missing.
 - **`tests/test-navigation-memory-events.cjs` `test1_enumCount`** now asserts a floor (`EVENT_TYPES.size >= 19`) rather than an exact count, so adding event types in later phases no longer breaks the Phase 109 baseline test (the required-types membership loop still pins the baseline). Resolves a deferred item from Phase 88.2.
@@ -111,7 +116,7 @@ Phase 95.6 -- install-cache Windows hardening + skill-loop resilience. Closes ca
 - **Explicit Brain / methodology access declared on all sub-agents (D-10)**, plus a **Deferred Tool Loading note** in `mcp-server-brain/CLAUDE.md` (D-11b: the <=10-15 startup-tools cap + schema-on-demand rule + the current Brain-MCP startup tool count of 6).
 - **`docs/install/PACKAGING-PATHS.md` (D-05e/f)** -- the four distribution paths (Marketplace + GitHub, Marketplace + npm, ZIP-URL, CI-Docker pre-bake) with "use this when..." guidance; the `CLAUDE_CODE_PLUGIN_SEED_DIR` / `CLAUDE_CODE_PLUGIN_CACHE_DIR` pre-bake flagged as the recommended NATO faculty deployment.
 - **README `## Manual Recovery` section (D-04)** -- step-by-step recovery for a halted install (re-run `install.sh`; manual agent symlink + `/mos:doctor --statusline-visibility --fix`; verify with `/mos:doctor --all`), plus a CMD-vs-PowerShell note. And a permission-prompts note in the README install section (D-03 ergonomics: 10+ prompts are normal, `always allow` shortcut, the new `WARN: skipping skill` line).
-- **Case #4 install-failure autopsy.** `docs/autopsies/2026-05-09-gary-laben-install-failure.md` -- the 2026-05-08/09 Gary Laben install (Surface Pro 7, Windows 11, the live 66-minute call), six root causes (Windows MAX_PATH, install.sh skill-loop halt, `@mindrian/os` never published, statusLine never registered, PowerShell-vs-CMD shell variability, permission-prompt fatigue), the install-cache failure family pattern (case #4 after #1 wrong-workspace, #2 Phase 93 drift-recovery, #3 Phase 95.2 atomic-recovery), the Phase 95.6 fixes, cross-references. Brain key UUIDs redacted per REC-05.
+- **Case #4 install-failure autopsy.** `[private case archive]` -- the 2026-05-08/09 the Wave-2 tester install (Surface Pro 7, Windows 11, the live 66-minute call), six root causes (Windows MAX_PATH, install.sh skill-loop halt, `@mindrian/os` never published, statusLine never registered, PowerShell-vs-CMD shell variability, permission-prompt fatigue), the install-cache failure family pattern (case #4 after #1 wrong-workspace, #2 Phase 93 drift-recovery, #3 Phase 95.2 atomic-recovery), the Phase 95.6 fixes, cross-references. Brain key UUIDs redacted per REC-05.
 - **`scripts/check-first-touch-drift.cjs` + `tests/test-first-touch-drift-scanner.cjs` (D-07)** -- scans the first-touch surfaces for em-dashes, stale version literals (scoped to greeting copy), and `BSL-1.1` adjacent to "open source" (SEED-007 pattern 3). `MOS_TEST_SCAN_DIR` override.
 
 ### Changed
@@ -328,7 +333,7 @@ Phase 117 was originally targeted at v1.13.0-beta.3 per CANON-PHASE-MAP.md, then
 ### Manual action items
 
 - **POST-MERGE WEBSITE EDIT:** apply the website hero rewrite from `docs/copy/115-website-hero.md` to `~/mindrian-website/[hero file]`. The website repo is independent of MindrianOS-Plugin; this is NOT auto-applied. The deliverable + step-by-step is in `docs/copy/115-website-hero.md`.
-- **VALIDATION WEEK:** dispatch `tests/fixtures/115-validation-email-template.md` to the 5-tester cohort (Lawrence Aronhime + Justin Stitzlein + Aryeh Holtzberg + Adam Peters + Shmuel Schuman) per D-13 (async, 48h reply window). Synthesize replies into `tests/fixtures/115-tester-rubric.md`.
+- **VALIDATION WEEK:** dispatch `tests/fixtures/115-validation-email-template.md` to the 5-tester cohort (Lawrence Aronhime + a tester + Aryeh Holtzberg + Adam Peters + a tester) per D-13 (async, 48h reply window). Synthesize replies into `tests/fixtures/115-tester-rubric.md`.
 - **D-20 ROLLBACK GATE (Pitfall 5 pre-commit):** if validation lands < 4-of-5 vivid recent memory, execute `tests/manual/115-rollback-procedure.md` step-by-step. Pre-committed; no live deliberation.
 - **MARKETPLACE Gate 5 (deferred):** ref-pin `~/mindrian-marketplace/.claude-plugin/marketplace.json` `source.ref` to `v1.13.0-beta.3` ONLY after the 5-tester async validation passes 4-of-5 AND the 3-tester live empathy audit (`tests/manual/115-acceptance.md`) reports 2/3 pass. Until then, Phase 115 ships as a LOCAL-ONLY tagged build (no `git push`, no marketplace ref-pin).
 
@@ -932,7 +937,7 @@ Stable users on 1.11.0 are not affected.
 
 ### Fixed (Phase 93 D1: Brain telemetry visibility)
 
-- **`mcp-server-brain/brain-admin.cjs`** column-name mismatch (5 occurrences across `cmdList` + `cmdUsage`): read `total_requests` and `last_request_at` instead of stale/dead `request_count` and `last_used_at`. Result: `/mos:admin keys` and `/mos:admin usage` now display real adoption numbers instead of universal zero. Verified post-fix: jsagir Desktop=378, Leah Aronhime=37, Lawrence Aronhime=26, plus six smaller users — matches Supabase ground truth.
+- **`mcp-server-brain/brain-admin.cjs`** column-name mismatch (5 occurrences across `cmdList` + `cmdUsage`): read `total_requests` and `last_request_at` instead of stale/dead `request_count` and `last_used_at`. Result: `/mos:admin keys` and `/mos:admin usage` now display real adoption numbers instead of universal zero. Verified post-fix: jsagir Desktop=378, an admin-key holder=37, Lawrence Aronhime=26, plus six smaller users — matches Supabase ground truth.
 - **`mcp-server-brain/lib/auth.cjs`** `logUsage()` insert column: write to `api_key` instead of nonexistent `key_id`. Previous code silently dropped 452 telemetry events with `PGRST204` errors swallowed by an upstream fire-and-forget `.catch()`. Brain usage log now fills correctly from this release forward; tool-level granularity restored.
 
 ### Added (Phase 93 D2: install-cache drift recovery)
@@ -1061,7 +1066,7 @@ The first command refreshes the marketplace catalog so v1.11.0 becomes visible.
 
 ## [1.11.0-beta.1] - 2026-04-27
 
-Beta release of the Reverse Salient (RS) Discovery Engine for opt-in testers (Justin / Aryeh). Stable users on v1.10.19 are NOT auto-updated; opt-in is explicit. Phase 91 Navigation Engine is NOT yet wired -- coming in beta.2. Tester sign-off promotes to stable v1.11.0 in Phase 91.5.
+Beta release of the Reverse Salient (RS) Discovery Engine for opt-in testers (the Wave-1 testers). Stable users on v1.10.19 are NOT auto-updated; opt-in is explicit. Phase 91 Navigation Engine is NOT yet wired -- coming in beta.2. Tester sign-off promotes to stable v1.11.0 in Phase 91.5.
 
 ### Tester Opt-In
 
@@ -2415,7 +2420,7 @@ Windows hotfix and Mac parity release. Ships Phase 85 (10 plans) addressing cros
 - **(WIN-FIX-F) run-hook.cmd exit code propagation (security-adjacent).** `hooks/run-hook.cmd` on Windows was swallowing bash exit codes because `%ERRORLEVEL%` inside an `if(...)` block is parse-time expanded, not runtime. PreToolUse write-scope-check returned 0 even when bash emitted exit 2, so the Phase 83 sealed-room write guard was silently inert on Windows for v1.10.7 and v1.10.8. **Security-adjacent: the sealed-room write guard was inert on Windows in v1.10.7 and v1.10.8. If you moved files into another room on Windows during that window, Larry's judgment was the only thing stopping it.** Fix uses `setlocal enabledelayedexpansion` with `!ERRORLEVEL!` captured into RC and `endlocal & exit /b %RC%` across all three bash invocation branches. Regression fixture at `tests/test-run-hook-cmd.cjs`.
 - **(WIN-FIX-B) `vunknown` banner on Windows.** `scripts/session-start` was reading plugin.json via `python3 -c "import json; json.load(...)"`. On Windows fresh installs, `python3` resolves to the Microsoft Store alias stub and silently exits non-zero, the `|| echo unknown` fallback fires, and users see `vunknown` in their banner instead of the real version. Now uses node via `lib/core/platform.cjs` `readPluginJsonVersion()`.
 - **(WIN-FIX-H) Cross-platform banner rendering and dispatch.** Introduced `lib/core/platform.cjs` centralizing OS detection, terminal code page handling, and hook script path resolution. Session-start banner now renders correctly on all platforms (UTF-8 box-drawing with ASCII fallback on non-UTF-8 terminals), statusline wrapper paths resolve through the helper, and python3 invocations have been audited across scripts/ with OS-aware gating.
-- **Mac `stat -c` portable fallback (LAWRENCE-001).** Confirmed that session-start, sentinel-health-check, on-task-complete, and post-compact use a `portable_stat_mtime` helper handling both GNU and BSD `stat`. Reported by Lawrence Aronhime (Johns Hopkins) via structured Mac environment audit on 2026-04-15.
+- **Mac `stat -c` portable fallback (LAWRENCE-001).** Confirmed that session-start, sentinel-health-check, on-task-complete, and post-compact use a `portable_stat_mtime` helper handling both GNU and BSD `stat`. Reported by Lawrence Aronhime via structured Mac environment audit on 2026-04-15.
 - **Lying header comment at `scripts/self-update` line 14** claimed the rewrite "abandons the atomic-swap-via-rename dance entirely" while line 325 still executed `mv`. Replaced with the truth: Windows uses `cp -a`, POSIX keeps `mv`, and the bootstrap handoff sidesteps the self-overwrite hazard.
 - **Regression fence:** new `tests/test-self-update-platform.cjs` covers the four scenarios from LASZLO-001 (win32 vs linux INSTALL_METHOD selection, `readPluginJsonVersion` helper without python3, `/tmp/` prefix resolution, and end-to-end bootstrap install in both branches) plus explicit J-1 ghost guards. Registered in `lib/memory/run-feynman-tests.cjs` (17/17 test files green).
 
@@ -2429,7 +2434,7 @@ Windows hotfix and Mac parity release. Ships Phase 85 (10 plans) addressing cros
 ### Credits
 
 - **László Személyi (Laszlo Szemelyi)**, Neumann Technology Platform, Hungary, for the detailed Windows self-update failure report including five screenshots of the `/mos:update` transcript (LASZLO-001, 2026-04-15). The "token-eating challenge" phrasing was the hook that surfaced the J family.
-- **Lawrence Aronhime**, Johns Hopkins, for the structured Mac environment audit covering nine sections from environment fingerprint to feature coverage analysis, including the Python ML dependency gap that drove the whitespace auto-install work (LAWRENCE-001, 2026-04-15).
+- **Lawrence Aronhime** for the structured Mac environment audit covering nine sections from environment fingerprint to feature coverage analysis, including the Python ML dependency gap that drove the whitespace auto-install work (LAWRENCE-001, 2026-04-15).
 
 ### Upgrade instructions
 
@@ -2520,7 +2525,7 @@ Upgrade path: standard two-command `/plugin marketplace update` followed by `cla
 
 ### Credit
 
-Bug reported by Lawrence Aronhime (lawrence@mindrian.ai, Prof., Johns Hopkins Carey Business School) on 2026-04-13 23:23. Lawrence has been running beta builds since v1.9.x and holds the lawrence@mindrian.ai admin Brain API key issued 2026-03-26. He built a same-night workaround on his own machine by injecting artifact data directly into `ROOM_DATA`, then filed the bug for the rest of the beta cohort. Eight releases shipped between his report and this fix. Thank you, Lawrence.
+Bug reported by Lawrence Aronhime (lawrence@mindrian.ai) on 2026-04-13 23:23. Lawrence has been running beta builds since v1.9.x and holds the lawrence@mindrian.ai admin Brain API key issued 2026-03-26. He built a same-night workaround on his own machine by injecting artifact data directly into `ROOM_DATA`, then filed the bug for the rest of the beta cohort. Eight releases shipped between his report and this fix. Thank you, Lawrence.
 
 ## [1.10.4] - 2026-04-14
 
@@ -2873,7 +2878,7 @@ onboard_steps:
 
 onboarding: true
 onboard_steps:
-  - "NEW: /mos:whitespace -- find what's MISSING in your venture. Maps gaps using embedding-space density analysis, based on Huan He's SemNovel research (Yale)."
+  - "NEW: /mos:whitespace -- find what's MISSING in your venture. Maps gaps using embedding-space density analysis, based on the researcher's SemNovel research (Yale)."
   - "MindrianOS now has a Model Data Room -- 168 artifacts across 10 sections, built from 45 meeting transcripts, 43 research papers, 35 PWS frameworks."
   - "HSI Spectral Analysis on real evidence -- 20 cross-domain innovation pairs discovered, reverse salients identified."