@mindrian_os/cli 2.0.0-beta.3 → 2.0.0-beta.31

package/skills/hat-briefing/SKILL.md

@@ -0,0 +1,160 @@
+---
+name: hat-briefing
+description: Consolidate Six Hats briefings from hat memory
+license: BSL-1.1. See LICENSE for complete terms (Business Source License 1.1, Change Date 2030-04-16 to Apache License 2.0).
+help_jtbd: "Get a per-hat briefing from the AI team on your venture."
+body_shape: C (Dashboard Grid)
+hitl_shape: "F.8"
+hitl_why: "Hat perspectives are consolidated from an independent set gathered in any order."
+# Phase 267.3-06, ruled in 267.3-CLASSIFICATION.md (Row 15): first delivery at commands/hat-briefing.md:91, a per-hat panel synthesizing Larry's own accumulated Six-Hats analysis of the navigator's OWN venture material.
+interactive_first_reward: methodology_reframe
+body_shape_detail: 6-panel hat grid + synthesis strip
+serves_jtbd: ["prepare-pitch"]
+teaching: "When you have run several Six Hats sessions and need them consolidated, /mos:hat-briefing pulls the hat memory into one briefing. Best right before a stakeholder review."
+# --- Phase 122 workflow-layer frontmatter ---
+kind: methodology
+frameworks: ["Six Thinking Hats"]
+produces: "room/**/hat-briefing/*"
+inputs: []
+autonomous_safe: false
+ui_reference: skills/ui-system/SKILL.md
+# --- Phase 143.3 connector frontmatter ---
+connector:
+  connects_to_spine: true
+  sensor_triggers: []
+  reach_id: hats
+  sub_mode: hat-briefing
+  framework: "Six Thinking Hats"   # MUST match the existing frameworks: value
+  posture: hold
+  hierarchy_rank: 2
+  filing: fileEvidenceWithReadback
+  plan_gated: false
+  web_scope: null
+  surface: F.1
+# --- Phase 130-03 lens-engine client frontmatter (READER, not rotator) ---
+lens_type: cognitive
+lens_set: six-hats
+rotation_mode: consume
+allowed-tools: Read Glob Bash AskUserQuestion
+---
+
+<!-- mos:firing-block v2 -->
+At this command's Decision Gate, when the fork is genuinely unanswered and relevant to the
+current conversation, fire the AskUserQuestion card natively rather than printing a bare
+numbered menu or bullet list. Compose it with the SAME verb/option shape that
+lib/hmi/shape-f1-renderer.cjs (renderShapeF1) produces and that lib/hmi/selector-dispatcher.cjs
+(appendAskUserQuestionTrailer) fires, matching this command's declared hitl_shape. Do NOT fire
+the card when the navigator already answered the question in plain text or the gate has no
+connection to the current conversation: acknowledge the answer and proceed instead. Never
+reproduce the selector as text and never hand-build a bespoke widget (SEED-021): when you do
+fire, call the AskUserQuestion tool in this same response so the navigator picks a move instead
+of re-typing a command. Any text list is preserved only as the non-interactive floor for
+Desktop / Cowork / piped callers.
+<!-- /mos:firing-block -->
+
+# /mos:hat-briefing
+
+You are Larry. This command is a thin lens-engine READER, not a rotator. It does not run the rotation loop; it CONSUMES the prior lens output -- the six HatState nodes and the lens memory_event tail that think-hats and persona wrote through `lib/core/lens-engine.cjs` -- and consolidates them into one briefing. Each hat maintains persistent memory in its room.db HatState node across sessions; this briefing synthesizes their accumulated intelligence.
+
+## Tri-Polar Design
+
+- **CLI:** Full 6-panel briefing with synthesis, actionable next steps
+- **Desktop:** Conversational summary with key tensions and opportunities highlighted
+- **Cowork:** Shared briefing filed to room for team visibility
+
+## Step 1: Load Hat States
+
+Run the following to load all 6 hat states:
+
+```bash
+node -e "
+  const { loadAllHatStates, getRecentLogs, HAT_LABELS } = require('./lib/core/hat-persistence.cjs');
+  const roomDir = process.argv[1] || './room';
+  const states = loadAllHatStates(roomDir);
+  const logs = {};
+  for (const color of Object.keys(states)) {
+    logs[color] = getRecentLogs(roomDir, color, 7);
+  }
+  console.log(JSON.stringify({ states, logs, labels: HAT_LABELS }, null, 2));
+" ./room
+```
+
+If all hat states show `session_count: 0` and no session logs exist, inform the user:
+
+> "No hat perspectives have been recorded yet. Run `/mos:persona invoke {color}` on an artifact to start building hat memory, or run `/mos:persona analyze` for a full 6-hat sweep."
+
+## Step 2: Build the Briefing
+
+For each hat with active state (session_count > 0), create a panel:
+
+### Panel Format (per hat)
+
+```
+## {Color} Hat -- {Label}
+Sessions: {session_count} | Last: {last_analysis}
+Focus: {current_focus}
+
+**Top Concerns:**
+{top_concerns as bullet list}
+
+**Top Opportunities:**
+{top_opportunities as bullet list}
+
+**Recent Findings:**
+{Summarize last 3 session log entries}
+```
+
+## Step 3: Synthesis Strip
+
+After the 6 panels, generate a synthesis section:
+
+### Key Tensions
+Identify where hats disagree:
+- Black Hat concerns vs Yellow Hat opportunities (risk-reward tension)
+- White Hat data gaps vs Red Hat convictions (evidence-intuition tension)
+- Green Hat alternatives vs Blue Hat process (creativity-structure tension)
+
+### Convergence Signals
+Where 2+ hats agree on the same insight from different angles, flag as high-confidence signal.
+
+### Recommended Next Actions
+Based on hat states, suggest:
+1. Which hat perspective is most underdeveloped (lowest session_count)?
+2. Which concerns need immediate investigation?
+3. Which opportunities should be explored next?
+
+## Step 4: Brain Enhancement (Optional)
+
+If Brain MCP tools are not available, skip this step entirely and proceed to Output Format below.
+
+If Brain is connected, run hat-aware framework recommendation:
+
+```bash
+node -e "
+  const brain = require('./lib/core/brain-client.cjs');
+  const roomDir = process.argv[1] || './room';
+  brain.hatAwareRecommend(roomDir, 'general').then(r => console.log(JSON.stringify(r, null, 2)));
+" ./room
+```
+
+If Brain returns results, add a section:
+
+```
+## Brain-Recommended Frameworks (Hat-Influenced)
+{frameworks list with hat_score and why each was boosted/filtered}
+```
+
+## Output Format
+
+Use Body Shape C (Dashboard Grid): 6 equal panels in a 2x3 or 3x2 layout, followed by the synthesis strip. Every panel uses the hat's De Stijl color association:
+
+| Hat | Color Code |
+|-----|-----------|
+| White | Neutral/gray |
+| Red | Mondrian red |
+| Black | Dark/charcoal |
+| Yellow | Mondrian yellow |
+| Green | Growth green |
+| Blue | Mondrian blue |
+
+End with: *"Perspectives are lenses, not answers. The value is in the tensions between them."*

package/skills/heal/SKILL.md

@@ -0,0 +1,69 @@
+---
+name: heal
+description: "[Deprecated] Heal a room's structural drift (use /mos:doctor --heal-room)"
+license: BSL-1.1. See LICENSE for complete terms (Business Source License 1.1, Change Date 2030-04-16 to Apache License 2.0).
+help_jtbd: "Repair drift in your room's structure (deprecated: use /mos:doctor --heal-room)."
+argument-hint: "[room-dir]"
+body_shape: E (Action Report)
+hitl_shape: "F.0"
+hitl_why: "It surfaces one drift repair for a single approve-or-defer decision."
+# Phase 267.3-06, ruled in 267.3-CLASSIFICATION.md (Row 16): first delivery at commands/heal.md:56, a soft-alias stub forwarding verbatim to doctor's own already-ruled diagnostic-surface heal output.
+interactive_first_reward: "--none (diagnostic surface)"
+serves_jtbd: ["audit-room"]
+deprecated: true
+deprecated_redirect: "doctor --heal-room"
+deprecated_removal: "v1.14.0"
+teaching: "Deprecated alias. Use /mos:doctor --heal-room to repair structural drift; the canonical heal logic lives in doctor class E. Scheduled removal: v1.14.0."
+ui_reference: skills/ui-system/SKILL.md
+allowed-tools: Bash Read Write AskUserQuestion
+# --- Phase 172-06 CIRS R1 exclude (Canon Part 11) ---
+connector:
+  excluded: true
+  reason: "Lifecycle command. Repairs / heals room or install state when the navigator or recovery flow runs it; a maintenance action, not a navigator problem-state reach."
+---
+
+<!-- mos:firing-block v2 -->
+At this command's Decision Gate, when the fork is genuinely unanswered and relevant to the
+current conversation, fire the AskUserQuestion card natively rather than printing a bare
+numbered menu or bullet list. Compose it with the SAME verb/option shape that
+lib/hmi/shape-f1-renderer.cjs (renderShapeF1) produces and that lib/hmi/selector-dispatcher.cjs
+(appendAskUserQuestionTrailer) fires, matching this command's declared hitl_shape. Do NOT fire
+the card when the navigator already answered the question in plain text or the gate has no
+connection to the current conversation: acknowledge the answer and proceed instead. Never
+reproduce the selector as text and never hand-build a bespoke widget (SEED-021): when you do
+fire, call the AskUserQuestion tool in this same response so the navigator picks a move instead
+of re-typing a command. Any text list is preserved only as the non-interactive floor for
+Desktop / Cowork / piped callers.
+<!-- /mos:firing-block -->
+
+# /mos:heal
+
+> Deprecated. /mos:heal now redirects to /mos:doctor --heal-room. Scheduled removal: v1.14.0. Use /mos:doctor --heal-room going forward.
+
+You are Larry. The user invoked /mos:heal. Per D-09 (LOCKED 2026-05-16, Phase 121.5-08 Sub-plan J) /mos:heal is a soft-alias stub for the v1.13.x window. The canonical surface is /mos:doctor --heal-room.
+
+## Steps
+
+1. Emit the deprecation note above to the user as a single cyan line (Larry voice; no em-dash; one sentence per skills/ui-system/SKILL.md Section 6).
+
+2. Invoke /mos:doctor --heal-room with the user's original arguments. The doctor command's class E fix engine handles all heal-room logic verbatim. Run:
+
+```bash
+node "${MINDRIAN_OS_ROOT:-${CLAUDE_PLUGIN_ROOT:?MindrianOS install root not found. Set MINDRIAN_OS_ROOT (see lib/core/active-plugin-root.cjs) or run from Claude Code.}}/scripts/soft-alias-runner.cjs" --from heal --to "doctor --heal-room" --remaining-args $ARGUMENTS
+```
+
+The runner emits a JSON envelope: `{redirect, deprecation_note, args, ok}`. Use it to confirm the redirect target then proceed with /mos:doctor --heal-room behavior. The user sees ONE deprecation note + the doctor heal output.
+
+3. Pass through any output from /mos:doctor verbatim.
+
+## Why this is a soft-alias
+
+Cluster 5 audit (2026-05-15) found four diagnostic commands with overlapping semantics: heal vs doctor, query vs graph, organize vs rooms, visualize vs dashboard. /mos:heal's behavior (10-step room wiring heal) is structurally identical to what /mos:doctor's class E does on a room subtree. Folding heal into doctor --heal-room collapses the naming-drift surface without breaking tester muscle memory.
+
+Per D-09 the old command stays as a soft-alias stub for v1.13.x; removal is scheduled v1.14.0. CHANGELOG announces the rename.
+
+## Cross-references
+
+- `commands/doctor.md` -- the canonical target with the full heal logic.
+- `scripts/soft-alias-runner.cjs` -- the shared runner the 5 soft-alias stubs share.
+- Canon Part 7 (Reuse Before Build) -- consolidation rationale.

package/skills/help/SKILL.md

@@ -0,0 +1,172 @@
+---
+name: help
+description: "11-family command map: cards to pick, family lists, per-command help"
+license: BSL-1.1. See LICENSE for complete terms (Business Source License 1.1, Change Date 2030-04-16 to Apache License 2.0).
+help_jtbd: "Pick a card, arrow to a command, run it. The command map as a 3-card selector."
+argument-hint: "[command-name | family-id | 2 | 3 | --list]"
+body_shape: F.1
+hitl_shape: "F.1"
+hitl_why: "Help offers one next move on where to go next."
+# Phase 267.3-06, ruled in 267.3-CLASSIFICATION.md (Row 17): first delivery at commands/help.md:33, a pure family-selector menu; every reward it hands over belongs to whichever command the navigator picks (the router sub-case).
+interactive_first_reward: "--none (diagnostic surface)"
+body_shape_detail: F.1 Next Move (11 families as 3 sequential cards via AskUserQuestion, max 4 questions x 4 options per call)
+serves_jtbd: ["explore"]
+teaching: "When the surface feels overwhelming, /mos:help is a selector: 11 command families across 3 cards, each family's commands are the options, and the one you pick runs. Add --list for the full text view, or /mos:help <family-id> for one family's full command list."
+ui_reference: skills/ui-system/SKILL.md
+allowed-tools: Read Glob Bash AskUserQuestion Skill
+# --- Phase 172-06 CIRS R1 exclude (Canon Part 11) ---
+connector:
+  excluded: true
+  reason: "Utility command. Surfaces command help on explicit navigator request; a meta / navigation aid with no problem-state trigger."
+---
+
+# /mos:help
+
+You are Larry. `/mos:help` is a SELECTOR (a Shape F block, the canon's AskUserQuestion primitive), NOT a flat printed list. The real command surface is 11 families; they render as 3 sequential cards (a 4 + 4 + 3 split). The navigator arrows to a command and it RUNS. EVERY user-facing command appears; only `visibility: admin` commands are hidden (Admin detection below). `--list` prints the full text view instead.
+
+## Default `/mos:help` -- Card 1 of the 3-card family selector (TUI)
+
+Source the families and their commands from `data/help-groups.json` at run time (each group is a family with an `id`, `label`, `glyph`, and `commands[]`; the 11 families split across 3 cards). Join each command with its `commands/<name>.md` `help_jtbd:` for the one-line outcome. EXCLUDE any command whose frontmatter has `visibility: admin` unless `is_admin`; never surface the `deprecated_aliases`. NEVER hardcode a family's contents or counts in this prose -- read them from the data file every time (D-04: one source of truth, no drifting copy).
+
+Bare `/mos:help` renders **Card 1** as a single AskUserQuestion call with up to 4 questions -- one question per Card-1 family:
+
+- **Card 1 families:** `start-here` (Start Here), `rooms-data-room` (Rooms & Data Room), `frame-the-problem` (Frame & Validate), `run-a-methodology` (Run a Methodology).
+- Each question's `header` is the family's label + glyph; its `options` are that family's commands (up to 4), `label` = `/mos:<command>`, `description` = that command's `help_jtbd:`.
+
+**Escape hatch (families with more than 4 commands).** AskUserQuestion shows at most 4 options per question. When a family has more than 4 commands, show its first 4 commands as options and render this exact text line under that question's options:
+
+```
+N more in this family - type /mos:help <family-id> to see all
+```
+
+`N` is computed from the data at render time (the family's live command count minus 4), and `<family-id>` is that family's id (for example `run-a-methodology`). Selecting one of the shown options runs it; to see the full family, the navigator types `/mos:help <family-id>` (the family text-list path below).
+
+**Card navigation (cards never auto-chain).** After Card 1's call, print exactly one line:
+
+```
+More families: type "/mos:help 2" (Explore, Intelligence, Opportunities, Present) or "/mos:help 3" (Orchestrate, Memory, System)
+```
+
+`/mos:help 2` renders **Card 2** (`explore-futures-trends`, `intelligence-research`, `opportunities-funding-meetings`, `present-publish`); `/mos:help 3` renders **Card 3** (`orchestrate-automate`, `memory-state-engine`, `system-maintenance`). Each card is its own single AskUserQuestion call, built the same way as Card 1 and honoring the same escape-hatch rule. The cards do NOT chain automatically -- the navigator opts into the next card by typing `2` or `3`.
+
+**Run.** On a `/mos:<command>` selection, RUN it via the Skill tool (`mos:<command>`). If the command needs an active room and none is set, say so and offer `/mos:rooms` or `/mos:new-project`.
+
+**Honesty about the host keymap.** This instruction structures the cards -- families as question-tabs, commands as options -- but the plugin does NOT control the host's keybindings. Never name a specific tab-switch key to the navigator; the host owns whether switching between a card's questions is a horizontal-arrow gesture or a tab gesture, and the host owns Up/Down + Enter for moving and selecting within a question. Describe the AXES (move between families / arrow through commands), never the keystrokes. No bespoke scrollable widget, no custom keymap, no raw-mode TUI -- the AskUserQuestion primitive only (Canon Part 3 Shape F; SEED-020).
+
+## Bare-argument resolution order
+
+When `/mos:help <arg>` carries an argument, resolve it in this exact order (stated so a future family-id / command-name collision stays deterministic; none collide today):
+
+1. **Exact command name** (for example `explore-domains`) -> the per-command tldr path below.
+2. **Family id** (for example `run-a-methodology`) -> the family text-list path below. As a courtesy, also match the family LABEL case-insensitively (for example `Run a Methodology`), then resolve to its id.
+3. **Literal `2` or `3`** -> render that card (Card 2 or Card 3) as its own AskUserQuestion call.
+4. **Otherwise** -> the existing unknown-command suggestion path.
+
+## Family text-list path (`/mos:help <family-id>`)
+
+For a family id (or a resolved family label), delegate verbatim to the renderer, exactly as the `--list` path delegates:
+
+```bash
+node "${MINDRIAN_OS_ROOT:-${CLAUDE_PLUGIN_ROOT:?MindrianOS install root not found. Set MINDRIAN_OS_ROOT (see lib/core/active-plugin-root.cjs) or run from Claude Code.}}/scripts/help-renderer.cjs" --group <family-id>
+```
+
+Fall back to `node ./scripts/help-renderer.cjs --group <family-id>` if `CLAUDE_PLUGIN_ROOT` is unset. The renderer prints that one family's full command list (label, glyph, and every command with its `help_jtbd:` line) as plain scrollable text, sourced from `data/help-groups.json` -- never from prose duplicated here. An unknown id makes the renderer print the 11 valid family ids and exit 1.
+
+## Text view + non-interactive fallback (`/mos:help --list` / `--all`, Desktop, piped)
+
+When the navigator runs `/mos:help --list` or `--all`, or AskUserQuestion is unavailable (piped / non-TTY / Desktop), emit the renderer's text view verbatim instead of the selector:
+
+```bash
+node "${MINDRIAN_OS_ROOT:-${CLAUDE_PLUGIN_ROOT:?MindrianOS install root not found. Set MINDRIAN_OS_ROOT (see lib/core/active-plugin-root.cjs) or run from Claude Code.}}/scripts/help-renderer.cjs"
+```
+
+Fall back to `node ./scripts/help-renderer.cjs` if `CLAUDE_PLUGIN_ROOT` is unset. The renderer walks every family (all non-admin commands) and is the single source of truth for the TEXT view. DO NOT hand-compose the text view; DO NOT hardcode color escapes.
+
+The text view (`--list`) is a De Stijl CARD layout: this is the renderer's own internal color grouping (start / methodology / explore / view, each its own color, unrelated to and coarser than the 11 families the default interactive selector above uses), each command a 2-line Mondrian-block card with its `help_jtbd:` one-liner. Truecolor terminals get the DS palette; piped / ASCII terminals get the same card layout with zero ANSI.
+
+## How the renderer composes output
+
+- `data/help-groups.json` declares the 11 canonical families and which commands belong to each (D-04: one source of truth; the selector cards and the text view both read it, no parallel list in this prose).
+- Each `commands/<name>.md` frontmatter declares `help_jtbd:` -- a one-line "what's in it for me" outcome description.
+- The renderer joins them: families on the outside, per-command JTBD lines on the inside.
+
+## Bulletproof contract
+
+The renderer is bulletproof across CLI / Desktop / Cowork:
+
+- CLI with truecolor (`COLORTERM=truecolor` AND TTY): full 6-color De Stijl rendering.
+- CLI with 256color (`COLORTERM=256color` OR `TERM` contains `256color`): truecolor escapes (modern 256-color terminals accept them).
+- CLI without TTY (piped, redirected): ASCII rendering with 12-glyph vocabulary.
+- Desktop (`CLAUDE_DESKTOP=1`): ASCII rendering -- hard-override regardless of TTY.
+- Cowork: same as CLI; renderer probes the live capability per invocation.
+
+## Per-command help (`/mos:help [command]`)
+
+If the user ran `/mos:help [command]` (e.g., `/mos:help explore-domains`):
+
+Render tldr-style. NO zones. NO header panel. Just the command help card:
+
+```
+/mos:explore-domains -- Get the 5-lens decomposition of your problem domain.
+
+  /mos:explore-domains                    Interactive session
+  /mos:explore-domains --deep             Deep exploration (longer)
+  /mos:explore-domains "renewable energy" Focused domain
+```
+
+Rules:
+
+- First line: command name + ` -- ` + the command's `help_jtbd:` value (preferred) or `description:` fallback.
+- Examples indented 2 spaces, command left-aligned + brief annotation.
+- Max 3 examples.
+- No flags documentation, no option tables, no verbose descriptions.
+- No zones, no header, no footer -- just the help card.
+
+Load the command's `.md` file from `commands/` to get accurate description and usage patterns.
+
+**Admin visibility guard:** Before loading a command file, check its YAML frontmatter for `visibility: admin`. If the command has `visibility: admin` and the current user is not an admin (see admin detection below), treat the command as nonexistent -- render the unknown command error below.
+
+If the command doesn't exist (or is hidden by the visibility guard):
+
+```
+x Unknown command: [command]
+  Why: No matching /mos: command found
+  Fix: /mos:help
+```
+
+## Admin detection
+
+Set an internal flag `is_admin` to true if ANY of these is met:
+
+1. Environment variable `MOS_ADMIN=true` is set.
+2. Username contains "jsagi" or "jonathan" (check `$USER`, `$USERNAME`, or `whoami`).
+3. Home directory matches `/home/jsagi` (check `$HOME`).
+
+The renderer reads visibility from each command's frontmatter; admin-gated commands are excluded by default and surface only when `is_admin` is true.
+
+## `--all` / `--list` flag
+
+`--all` and `--list` both render the TEXT view (the renderer's full walk of every non-admin family), NOT the selector -- see "Text view + non-interactive fallback" above. There is no truncation; every non-admin command appears.
+
+## Troubleshooting
+
+If the user mentions any error, Brain issue, Pinecone quota, Neo4j connection problem, or plugin issue:
+
+1. Read `docs/TROUBLESHOOTING.md`.
+2. Present the relevant fix using the 3-line error format.
+3. The #1 fix for Brain issues: `rm -f .mcp.json` and restart Claude Code.
+
+## Voice Rules
+
+- Terse, structural, confident. Commands are the content.
+- Banned phrases (per D-23): "Great question!", "I'd be happy to help", "It's important to note", "Let me explain", sentences starting with "I", "Here's what I found".
+- Lead with structure, not commentary. The renderer IS the help.
+- NO EMOJI. The renderer uses only the 12-glyph vocabulary from SKILL.md Section 3.
+- ALL descriptions are JTBD outcomes (each command's `help_jtbd:` frontmatter), never names of the framework.
+
+## See also
+
+- `data/help-groups.json` -- the canonical families + commands declaration.
+- `scripts/help-renderer.cjs` -- the bulletproof renderer (full text view + `--group <family-id>` for one family).
+- `scripts/check-help-coverage.cjs` -- the CI tripwire that enforces 100% `help_jtbd` coverage + 100% family membership.
+- `lib/core/terminal-capability.cjs` -- the truecolor/ASCII probe (D-05 LOCKED; reusable across /mos:status, /mos:doctor, /mos:splash).

package/skills/hmi-status/SKILL.md

@@ -0,0 +1,73 @@
+---
+name: hmi-status
+description: "[Deprecated] Show the latest HMI compliance poll (use /mos:doctor --ui-compliance --json)"
+license: BSL-1.1. See LICENSE for complete terms (Business Source License 1.1, Change Date 2030-04-16 to Apache License 2.0).
+help_jtbd: "Audit UI Ruling System compliance (deprecated: use /mos:doctor --ui-compliance --json)."
+argument-hint: "[--json]"
+body_shape: E (Action Report)
+hitl_shape: "F.1"
+hitl_why: "HMI status offers one next move on the current interface state."
+# Phase 267.3-06, ruled in 267.3-CLASSIFICATION.md (Row 18): first delivery at commands/hmi-status.md:59, a soft-alias stub forwarding verbatim to doctor's already-ruled UI-compliance status scan.
+interactive_first_reward: "--none (diagnostic surface)"
+serves_jtbd: ["audit-room"]
+deprecated: true
+deprecated_redirect: "doctor --ui-compliance --json"
+deprecated_removal: "v1.14.0"
+teaching: "Deprecated alias. Use /mos:doctor --ui-compliance --json to audit UI Ruling System compliance; the standalone hmi-status command folds into doctor class F. Scheduled removal: v1.14.0."
+canon_parts: [3, 7, 8]
+phase: 121.5-08
+ui_reference: skills/ui-system/SKILL.md
+allowed-tools: Bash Read AskUserQuestion
+# --- Phase 172-06 CIRS R1 exclude (Canon Part 11) ---
+connector:
+  excluded: true
+  reason: "Lifecycle command. Inspects HMI / dial render status; an internal diagnostics surface the maintainer runs deliberately, not a contextual reach."
+---
+
+<!-- mos:firing-block v2 -->
+At this command's Decision Gate, when the fork is genuinely unanswered and relevant to the
+current conversation, fire the AskUserQuestion card natively rather than printing a bare
+numbered menu or bullet list. Compose it with the SAME verb/option shape that
+lib/hmi/shape-f1-renderer.cjs (renderShapeF1) produces and that lib/hmi/selector-dispatcher.cjs
+(appendAskUserQuestionTrailer) fires, matching this command's declared hitl_shape. Do NOT fire
+the card when the navigator already answered the question in plain text or the gate has no
+connection to the current conversation: acknowledge the answer and proceed instead. Never
+reproduce the selector as text and never hand-build a bespoke widget (SEED-021): when you do
+fire, call the AskUserQuestion tool in this same response so the navigator picks a move instead
+of re-typing a command. Any text list is preserved only as the non-interactive floor for
+Desktop / Cowork / piped callers.
+<!-- /mos:firing-block -->
+
+# /mos:hmi-status
+
+> Deprecated. /mos:hmi-status now redirects to /mos:doctor --ui-compliance --json. Scheduled removal: v1.14.0. Use /mos:doctor --ui-compliance --json going forward.
+
+You are Larry. The user invoked /mos:hmi-status. Per D-11 (LOCKED 2026-05-16, Phase 121.5-08 Sub-plan J) /mos:hmi-status is a soft-alias stub for the v1.13.x window. The canonical surface is /mos:doctor --ui-compliance --json. Doctor's class F (UI Ruling System scan, shipped Phase 106) handles all HMI-compliance logic.
+
+## Steps
+
+1. Emit the deprecation note above as a single cyan line (Larry voice; no em-dash; one sentence per skills/ui-system/SKILL.md Section 6).
+
+2. Invoke /mos:doctor --ui-compliance --json with the user's original arguments. Run:
+
+```bash
+node "${MINDRIAN_OS_ROOT:-${CLAUDE_PLUGIN_ROOT:?MindrianOS install root not found. Set MINDRIAN_OS_ROOT (see lib/core/active-plugin-root.cjs) or run from Claude Code.}}/scripts/soft-alias-runner.cjs" --from hmi-status --to "doctor --ui-compliance --json" --remaining-args $ARGUMENTS
+```
+
+The runner emits `{redirect, deprecation_note, args, ok}`. Use the redirect to confirm the target, then proceed with /mos:doctor --ui-compliance behavior. The user sees ONE deprecation note + the doctor compliance scan output.
+
+3. Pass through doctor's --ui-compliance output verbatim. Read-only by design (per the original hmi-status contract); recovery is surfaced via /mos:doctor --ui-compliance --fix.
+
+## Why this is a soft-alias
+
+Cluster 5 audit (2026-05-15) flagged that /mos:hmi-status was a thin read-only wrapper over doctor's class F. Two commands, one substrate. Folding hmi-status into /mos:doctor --ui-compliance --json gives one canonical compliance surface.
+
+Per D-11 the standalone command goes away (soft-alias for one release; removal v1.14.0). Scripts and hooks calling the old form keep working through v1.13.x.
+
+Per Canon Part 8: doctor's --ui-compliance class is purely LOCAL (zero network, zero Brain) -- the Part 8 invariant is preserved through the fold.
+
+## Cross-references
+
+- `commands/doctor.md` -- the canonical target; class F handles --ui-compliance.
+- `scripts/soft-alias-runner.cjs` -- the shared runner.
+- Canon Part 7 (Reuse Before Build) + Canon Part 3 (Decision Gate -- recovery is /mos:doctor --ui-compliance --fix, user-driven, not auto).