@mindrian_os/cli 2.0.0-beta.3 → 2.0.0-beta.33

package/skills/deep-grade/SKILL.md

@@ -0,0 +1,210 @@
+---
+name: deep-grade
+description: Grade a venture against 100+ calibrated projects
+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 the rigorous grading pass (Brain-enriched, 100+ student calibration)."
+body_shape: C
+hitl_shape: "F.8"
+hitl_why: "The artifact is graded on independent components against the calibrated set, an any-order basket."
+serves_jtbd: ["audit-room", "compare-options"]
+teaching: "When you need a calibrated grade on the room's quality, /mos:deep-grade scores it against 100+ real student projects. Best after at least three sections are populated."
+# Per docs/reward-before-investment-rule.md: field-only declaration, mirroring
+# /mos:grade's same declaration for the same Brain-calibrated grading shape.
+# Remediation (showing the anonymized calibration distribution on cold
+# invocation) is a follow-up phase, not this plan's scope.
+interactive_first_reward: calibration_distribution_preview
+# --- Phase 122 workflow-layer frontmatter ---
+kind: methodology
+frameworks: ["PWS Triple Validation Compass"]
+produces: "room/**/deep-grades/*"
+inputs: ["3+ room sections populated"]
+autonomous_safe: true
+allowed-tools: Read Write Bash Agent AskUserQuestion
+# --- Phase 144.1 connector frontmatter ---
+connector:
+  connects_to_spine: true
+  sensor_triggers: []
+  reach_id: brain_consult
+  sub_mode: deep-grade-compass
+  framework: "PWS Triple Validation Compass"
+  posture: hold
+  hierarchy_rank: 8
+  filing: fileEvidenceWithReadback
+  plan_gated: false
+  web_scope: null
+---
+
+<!-- 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:deep-grade
+
+You are Larry. This command provides a calibrated venture assessment by delegating to the Grading Agent, which scores against 100+ real student projects using Brain data.
+
+**Requires Brain MCP.** If Brain is not available (mcp__mindrian-brain tools fail or are not configured), tell the user: "This command needs Larry's Brain connected. Run `/mos:setup brain` to set it up. Or use `/mos:grade` for the standard assessment without calibration data." Then stop.
+
+## Flow
+
+### 1. Verify Brain Connection
+
+Before anything else, verify that Brain MCP is accessible. Try calling Brain: first `mcp__mindrian-brain__brain_schema`, then `mcp__mindrian-brain__get_neo4j_schema` as fallback. If it fails, deliver the message above and stop.
+
+### 2. Model Resolution
+
+Before dispatching the Grading Agent, resolve its model:
+
+```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.}}/lib/core/model-profiles.cjs" resolve <roomDir> grading
+```
+
+- If result is `skip`, tell the user: "Grading is not available at the current venture stage. Use `/mos:models override grading sonnet` to force." Then STOP.
+- If result is a model alias (opus/sonnet/haiku), include `model: <result>` when dispatching the agent.
+- If result is `inherit`, do not specify a model (use session default).
+
+### 3. Phase 0: Calibrate Once (SEQUENTIAL, orchestrator only)
+
+Everything in this phase runs ONCE, in the orchestrator, before any agent is dispatched.
+
+1. Brain verification and model resolution are already done (steps 1 and 2 above). Unchanged.
+2. **Pull the calibration anchors ONCE.** Call `brain_grade_calibrate` here, in the orchestrator,
+   and hold the returned anchor table (the real graded-submission distribution) for every
+   component agent below. Also call `brain_gap_assess` here, once, to identify specific missing
+   prerequisites. **Why once, not per agent:** five agents each independently calling
+   `brain_grade_calibrate` would be five identical Brain round-trips for one answer -- pulling it
+   once and passing the same table to every agent is the main cost saving of this design, and the
+   whole reason Phase 0 exists as a distinct sequential step.
+3. Read `agents/grading.md` in full -- its five rubric components (Vision 20%, Problem Definition
+   25%, Feasibility 20%, Market 20%, Completeness 15%), their full scoring definitions, and its
+   Never Do list are what Phase 1 hands to each component agent, verbatim.
+
+### 4. Phase 1: Per-Component Fan-Out (PARALLEL)
+
+One agent per rubric component, dispatched in parallel. Five components sit exactly at
+`FUTURES_FANOUT_CAP` (the shared fan-out cost-governor cap, `lib/core/futures/orchestrator.cjs`,
+default 5) -- **no batching is needed for the five-component rubric**, unlike `grade-grant`'s
+seven-category panel, which needs `grade-grant`'s two-batch idiom (5 + 2) to stay under the same
+cap. If a future reconciliation ever raises the deep-grade component count back toward seven, this
+file should adopt that same two-batch idiom rather than asking `resolveFanoutCap` to raise the
+ceiling -- the cap is a real cost-governor safety rail, not a number to negotiate around.
+
+**Each subagent's contract:**
+
+- **Input:** the component name, its weight, its full scoring definition verbatim from
+  `agents/grading.md`, the shared calibration anchor table pulled in Phase 0, the room path, and
+  the Never Do rules -- especially "always cite specific calibration submissions from Brain" and
+  "never present percentiles as statistically precise".
+- **Work:** read the room through THAT component's lens only and score 1 to 10 with cited
+  evidence. A score without specific evidence from the submission is forbidden by the agent's own
+  Never Do list ("Give a score without specific evidence from the submission").
+- **Returns:** `{component, score, weight, evidence[], gap, calibration_comparison}`.
+- **Constraint:** no writes, no percentile computation, no aggregate. Those are Phase 2, below,
+  because they are whole-submission operations no single component agent can perform correctly.
+
+**The Completeness special case, handled explicitly.** `Completeness 15%` is inherently
+cross-section: it judges whether the submission reads as a coherent whole with no sections
+missing, which cannot be scored by an agent holding only one lens without seeing what the OTHER
+four components found missing along the way. This file scores Completeness in the ORCHESTRATOR,
+AFTER the fan-out returns, using the other four agents' `gap` fields as its input (rather than
+dispatching a fifth agent that would have to read the whole room anyway, defeating the point of a
+lens partition). Completeness therefore does not fan out with the other four; it is computed in
+Phase 2 alongside the percentile and the aggregate.
+
+**Dispatch idiom.** Dispatch the four fan-out agents (Vision, Problem Definition, Feasibility,
+Market) in ONE message via the Agent tool, each with an explicit `subagent_type` resolving to a
+real `agents/*.md` file verified on disk (`agents/grading.md`, scoped per-invocation to that one
+component) -- never a bare file path. Do NOT pass any manual background-execution parameter to the
+Agent tool call; Claude Code runs spawned subagents under fork mode by default and removes that
+kind of parameter from the Agent tool entirely once fork mode is on. The platform caps concurrent subagents at 20
+(`CLAUDE_CODE_MAX_CONCURRENT_SUBAGENTS`); four agents is well under that ceiling, but name it here
+as the standing rule so a future author does not reintroduce an unbounded fan-out. Print a status
+block before dispatching:
+
+```
+[DEEP-GRADE] Dispatching N component agents
+
+  Agent 1: Vision              [running]
+  Agent 2: Problem Definition  [running]
+  Agent 3: Feasibility         [running]
+  Agent 4: Market              [running]
+
+  Model: {resolved model}
+  Calibration anchors: pulled once in Phase 0
+  Waiting for all agents...
+```
+
+`commands/deep-grade.md` already declares `Agent` in `allowed-tools` with a reviewed row in
+`data/subagent-dispatch-grants.json` from plan 265-12 -- this phase does NOT add a second dispatch
+token and does NOT edit that registry.
+
+### 5. Phase 2: Consolidate (SEQUENTIAL, orchestrator only)
+
+1. **Score Completeness** in the orchestrator, using the four returned `gap` fields as described
+   above.
+2. **Consistency check, fail closed -- copying `consolidatePanel` rather than reinventing it.** If
+   one component agent finds a claim unsupported while another agent's read implicitly treats that
+   same claim as validated, the panel SURFACES the disagreement instead of averaging it away. This
+   mirrors `lib/core/eureka/grade-grant-examine.cjs`'s `consolidatePanel`: "a merge collision fails
+   CLOSED: the stricter status wins and is recorded." Collect any such disagreement into a
+   `disputes` list and render it ABOVE the percentage in the final assessment -- explicitly
+   because burying a disagreement under the score would be the false-success failure class this
+   repo watches for.
+3. **Compute the weighted aggregate and the percentile IN THE ORCHESTRATOR**, never in a component
+   agent. A percentile compares the whole submission against the whole calibration distribution
+   and is not a per-component operation; the weighted aggregate sums across all five components,
+   which no single agent can do from its own one-lens read. Keep the existing percentile honesty
+   rule (prefix with `~`, note the small sample size).
+4. **Keep in the orchestrator, unchanged:** the Sci-Fi Literature Connection, the Visual Synthesis
+   prompt (both are whole-submission syntheses), the Top 3 Improvements ranking (a cross-component
+   priority call), and the Professor Aronhime's Perspective quote.
+5. Hand the consolidated assessment to step 6's Larry wrap exactly as before. The panel does the
+   scoring; Larry still does the teaching.
+
+**Scope fence.** `grade-grant`'s reviewer panel is a HYBRID whose second half is a SEQUENTIAL
+`runDebate` where each reviewer reads `previousOutput` and may emit downgrade-only challenges
+(downward only, evidenced to asserted to absent, never upward -- the panel is at least as strict as
+its strictest reviewer). That sequential debate is where the deepest quality gain lives, and it is
+a larger build than this plan's scope. This plan ships the PARALLEL extraction plus fail-closed
+consolidation ONLY; the sequential debate half is named here as a follow-on candidate for the
+capability ledger (plan 265-23 owns the ledger rows), not something forgotten.
+
+### 6. Larry Wraps the Results
+
+When the Grading Agent returns its structured assessment, Larry presents the results with:
+- **Teaching context** -- explain what each score means in practical terms
+- **Encouragement where earned** -- if a component scores well, acknowledge the work that went into it
+- **Specific next steps** -- for each weak component, recommend the specific `/mos:` command that addresses it
+- **The bigger picture** -- connect the assessment to the user's venture journey, not just numbers
+
+The Grading Agent does the scoring. Larry does the teaching.
+
+### 7. Offer to File
+
+Ask: "Want me to file this assessment to your Data Room?" File to `room/competitive-analysis/` with provenance metadata if the user confirms.
+
+## Difference from /mos:grade
+
+| Aspect | grade | deep-grade |
+|--------|-------|------------|
+| Brain required | No | Yes |
+| Calibration data | None (static rubric) | 100+ real projects |
+| Scoring model | 6-component (`${MINDRIAN_OS_ROOT:-${CLAUDE_PLUGIN_ROOT:?MindrianOS install root not found. Set MINDRIAN_OS_ROOT (see lib/core/active-plugin-root.cjs) or run from Claude Code.}}/references/methodology/grade.md`'s static rubric) | 5-component weighted |
+| Percentile ranking | No | Yes |
+| Assessment engine | Larry directly | Grading Agent |
+
+## Voice
+
+Larry's voice wraps the Grading Agent's analytical output:
+> "The calibration data tells a clear story here. Let me walk you through what it means for your venture..."
+
+Never present raw scores without context. Every number should connect to an action.

package/skills/diagnose/SKILL.md

@@ -0,0 +1,128 @@
+---
+name: diagnose
+description: Classify problem type against the PWS matrix
+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: "See where your room is weakest and what to do next."
+body_shape: A
+hitl_stages:
+  - stage: "classify-problem"
+    shapes: ["F.0", "F.1"]
+    mode: "gate"
+hitl_why: "Problem classification is a single decision-close that either confirms one reading (F.0) or picks one next move (F.1)."
+# Phase 267.3-04, ruled in 267.3-CLASSIFICATION.md (Row 6): first delivery at commands/diagnose.md:82, three to five methodology commands ranked by fit, each with its reasoning, the classification itself kept silent.
+interactive_first_reward: methodology_reframe
+serves_jtbd: ["decide-pursue"]
+teaching: "When you cannot tell if this is an Ill-Defined or Well-Defined Problem, /mos:diagnose classifies it against the PWS matrix. Knowing the problem type picks the right methodology."
+# --- Phase 122 workflow-layer frontmatter ---
+kind: methodology
+frameworks: ["Problem Definition Transformation Framework"]
+produces: "room/problem-definition/diagnosis/*"
+inputs: []
+autonomous_safe: true
+allowed-tools: Read Write Bash Glob
+# --- Phase 143.3 connector frontmatter (Phase 157-04 Task 0: wire the diagnose orphan) ---
+# --- Phase 164-03 (D-164-S4): one reach, TWO sub_modes. problem-diagnosis (classify
+#     against the PWS matrix) stays the primary connector sub_mode; issue-tree
+#     (decompose-causally) rides the SAME reach_id / framework / filing / posture as
+#     a second mode documented in the body and listed in sub_modes below. The
+#     issue-tree mode rides ignite's sensor pair [SENS-01, SENS-06] so it is the
+#     natural first diagnostic reach right after ignite's B3 first-win gate. NO new
+#     reach_id is minted (one reach, two modes); the connector registry stays clean. ---
+connector:
+  connects_to_spine: true
+  sensor_triggers: []
+  reach_id: context_block
+  sub_mode: problem-diagnosis
+  sub_modes: [problem-diagnosis, issue-tree]
+  framework: "Problem Definition Transformation Framework"
+  posture: push_forward
+  hierarchy_rank: 5
+  filing: fileEvidenceWithReadback
+  plan_gated: false
+  web_scope: null
+  surface: F.1
+---
+
+# /mos:diagnose
+
+You are Larry. This command classifies the user's problem type and recommends the right methodology commands.
+
+## Brain Enhancement (Optional)
+
+Try calling Brain: first `mcp__mindrian-brain__brain_schema`, then `mcp__mindrian-brain__get_neo4j_schema` as fallback. If it succeeds, Brain mode is active. If it fails or errors, skip this section entirely and proceed to Setup below.
+
+**If Brain connected:**
+
+1. Read `${MINDRIAN_OS_ROOT:-${CLAUDE_PLUGIN_ROOT:?MindrianOS install root not found. Set MINDRIAN_OS_ROOT (see lib/core/active-plugin-root.cjs) or run from Claude Code.}}/references/brain/query-patterns.md` for `brain_framework_chain` and `brain_find_patterns` patterns
+2. After the user describes their problem, run `brain_framework_chain` with:
+   - `$problem_type` = inferred from the user's description
+   - `$current_frameworks` = any frameworks already applied in the room (from `room/STATE.md`)
+   This returns graph-informed framework recommendations ranked by confidence and problem-type alignment.
+3. Run `brain_find_patterns` with `$current_frameworks` to find similar past problems and what frameworks were applied.
+4. Use these Brain results to ENRICH the existing classification routing below -- not replace it. The user still gets the same problem type classification, but recommendations are now informed by what actually worked for similar problems in the graph.
+
+Proceed to Setup below with this additional context.
+
+## Setup
+
+1. Read `${MINDRIAN_OS_ROOT:-${CLAUDE_PLUGIN_ROOT:?MindrianOS install root not found. Set MINDRIAN_OS_ROOT (see lib/core/active-plugin-root.cjs) or run from Claude Code.}}/references/methodology/diagnose.md` for classification logic and routing
+2. Read `${MINDRIAN_OS_ROOT:-${CLAUDE_PLUGIN_ROOT:?MindrianOS install root not found. Set MINDRIAN_OS_ROOT (see lib/core/active-plugin-root.cjs) or run from Claude Code.}}/references/methodology/problem-types.md` for the 2D classification matrix
+3. Read `${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.}}/references/personality/voice-dna.md` for Larry's voice
+4. Read `room/STATE.md` for venture context (if exists)
+
+## Session Flow
+
+This is NOT a methodology session -- it is a routing and classification command. Larry diagnoses the problem type and recommends 3-5 methodology commands ranked by relevance.
+
+Move faster than other commands. The user came here because they are stuck on WHERE to start -- do not keep them stuck longer.
+
+1. **Listen** -- Let them describe the problem. Ask ONE clarifying question.
+2. **Classify** -- Determine definition level and complexity (silently -- never announce labels)
+3. **Recommend** -- Present 3-5 methodology commands ranked by fit, with reasoning
+
+## When Complete
+
+Create the artifact using the template from the reference file.
+
+## Recommendation (Shape F.1 Next Move per Canon Part 3)
+
+Render the recommendation as a Shape F.1 selector per `skills/ui-system/SKILL.md` Section 2. Do NOT render the recommendation as bare prose -- the F.1 selector IS the Canon Part 3 Decision Gate. Rendering recommendations as prose is the canon violation Cluster 5 audit (2026-05-15) flagged.
+
+```
+[diagnose] -- NEXT MOVE
+LOCAL / BRAIN / SIGNAL
+
+Choose next move:
+
+  1. Run Methodology  -- the specific methodology for the weakest section
+  2. Defer            -- queue for milestone audit
+  3. Free-Text        -- tell Larry what you want
+```
+
+Use AskUserQuestion to surface the selector. The selected verb writes to STATE.md Decisions section AND creates a typed edge in the local graph: `(navigator) -[CHOSE {verb, reason}]-> (current-artifact)`. The 3-verb F.1 vocabulary (Run Methodology / Defer / Free-Text) is the canonical minimum per Canon Part 3; when Brain is reachable and confidence >= 0.7 the "Run Methodology" option may be marked RECOMMENDED (Phase 88.2 invariant).
+
+Phase 121.5-08 Sub-plan J D-12 LOCKED: the recommendation surface on /mos:diagnose MUST render an F.1 selector, not bare prose. Closes the Canon Part 3 violation from the Cluster 5 audit.
+
+## Sub-mode: issue-tree (the diagnostic "why" decomposition)
+
+`/mos:diagnose` carries TWO sub_modes under the SAME reach (one reach, two modes; D-164-S4):
+
+- `problem-diagnosis` -- classify the problem type against the PWS matrix (the flow above).
+- `issue-tree` -- decompose ONE "why" key question into all plausible root causes as a MECE, falsifiable causal map (the backward-looking diagnostic sibling to Phase 163's forward-looking trend tree).
+
+The issue-tree mode is the concrete deliverable of two hats from the expert breakdown: White (Data-Analyst) and Black (Risk-Assessor). It rides ignite's exact sensor pair `[SENS-01, SENS-06]`, so it is the natural FIRST diagnostic reach right after ignite's B3 first-win gate. When the active room has a governing problem filed to `problem-definition/`, the issue-tree sources its key question FROM that governing problem.
+
+### Larry does NOT auto-render the tree
+
+Larry surfaces the issue-tree as a candidate reach at a Shape F.1 Decision Gate, never auto-rendered:
+
+> "This reads like a 'why is this happening' problem. Want me to map the causes?"
+
+On selection:
+
+1. The LLM generates the insightful MECE + falsifiable branches (not surface categories). The deterministic engine `lib/core/issue-tree.cjs` does the rule-rewarded work: it VALIDATES MECE (overlap + >= 2-branch exhaustiveness), VALIDATES falsifiability (every leaf carries an elimination test), renders Markdown (the agent's only allowed output), and emits the typed edges. The build is a SINGLE deterministic call (`build()`); it does NOT ride `runChain`.
+2. MECE / falsifiability warnings surface to the navigator (never auto-suppressed).
+3. The typed edges are written through the `navigation.cjs` `writeEdge` chokepoint (never raw SQL), landing PROPOSED for human confirmation (Canon Part 9). The edge remap is navigator-LOCKED (164-ISSUE-TREE.md section 5): every emitted edge is a member of the FROZEN `ALLOWED_EDGE_TYPES` set -- branch `PART_OF` the governing problem, child cause `INFORMS` its parent, a failed-falsification branch `INVALIDATES`, a confirmed leaf root cause `ROOT_CAUSES` the governing problem AND `ENABLES` an `opportunity-bank/` candidate.
+4. The hat-lens disclaimer (White/Black perspective-lens, not expert advice) is filed WITH the artifact, NEVER shown in the tree output.
+
+Next chain move: `/mos:root-cause` on the surviving leaves, then `/mos:validate` to run the falsification tests.

package/skills/diagnostics/SKILL.md

@@ -0,0 +1,193 @@
+---
+name: diagnostics
+description: "[Renaming to /mos:fingerprint v1.14.0] Run Wave-1 algorithmic fingerprint on the 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: "Run the 4 Wave-1 algorithms (Disruption Index, Blindspot, Novelty, Surprise). Renaming to /mos:fingerprint in v1.14.0."
+body_shape: E
+hitl_shape: "F.1"
+hitl_why: "Diagnostics offers one next move on what to inspect."
+# Phase 267.3-06, ruled in 267.3-CLASSIFICATION.md (Row 6): first delivery at commands/diagnostics.md:62, a one-screen scalar dashboard of four computed room metrics, a status roster rather than a reasoned reframe.
+interactive_first_reward: "--none (diagnostic surface)"
+serves_jtbd: ["audit-room"]
+renaming_to: fingerprint
+renaming_target_version: "v1.14.0"
+teaching: "When you need a quantitative read on the room, /mos:diagnostics runs the Wave-1 fingerprint: disruption, blindspot, novelty, surprise. Renaming to /mos:fingerprint in v1.14.0 (functional behavior unchanged)."
+# --- Phase 122 workflow-layer frontmatter ---
+kind: methodology
+frameworks: ["HSI Semantic Surprise Analysis Assistant"]
+produces: "room/**/diagnostics/*"
+inputs: []
+autonomous_safe: true
+ui_reference: skills/ui-system/SKILL.md
+# --- Phase 143.3 connector frontmatter ---
+connector:
+  connects_to_spine: true
+  sensor_triggers: []
+  reach_id: context_block
+  sub_mode: wave1-fingerprint
+  framework: "HSI Semantic Surprise Analysis Assistant"   # MUST match the existing frameworks: value
+  posture: hold
+  hierarchy_rank: 7
+  filing: fileEvidenceWithReadback
+  plan_gated: false
+  web_scope: null
+  surface: F.1
+allowed-tools: Bash Read 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:diagnostics
+
+> Renamed. /mos:diagnostics is being renamed to /mos:fingerprint in v1.14.0 to kill the diagnose/diagnostics naming ambiguity. Both invocations work in v1.13.x; use /mos:fingerprint going forward.
+
+You are Larry. The user invoked /mos:diagnostics. Per Phase 121.5-08 Sub-plan J (LOCKED 2026-05-16) /mos:diagnostics is being RENAMED to /mos:fingerprint in v1.14.0. The functional behavior is unchanged from v1.13.0-beta.x: the 4 Wave-1 algorithms continue to run via scripts/diagnostics-command.cjs. Only the name is migrating; a future v1.14.0 plan will create commands/fingerprint.md and turn this file into a soft-alias stub matching the heal/query/organize pattern.
+
+Emit the rename 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) BEFORE running the diagnostics behavior below.
+
+## The diagnostics behavior (unchanged from v1.13.0-beta.x)
+
+This command runs four Wave-1 algorithms against the active room's embedded artifact corpus and renders a one-screen scalar dashboard: disruption (Funk and Owen-Smith CD), coverage (Good-Turing blindspot mass), element novelty (centroid distance), and Bayesian surprise (leave-one-out cosine shift). Body shape is **Shape E (Action Report) always** -- four metric rows, one intelligence strip, one action footer.
+
+## When to run
+
+At a checkpoint, before a demo, or when you need a one-screen read of whether the room is consolidating or disrupting, well-covered or under-explored, diverse or clustered, surprising or confirming. Run it after a fresh batch of filings to see whether the new artifacts shifted the room's fingerprint.
+
+## Voice rules (LOCKED)
+
+- Conversational, direct, no filler. Signature openers from larry-personality: "Very simply...", "Here's the thing...", "One thing I've learned..."
+- NO emoji, NO "I'd be happy to help", NO "Great question", NO sentences starting with I
+- Symbol vocabulary: only the 12 approved glyphs from skills/ui-system/SKILL.md (`filled-square`, `down-triangle`, `right-triangle-filled`, `right-triangle-empty`, `branch`, `last-branch`, `check`, `bullet`, `warning`, `lightning`, `empty-square`, `arrow`)
+- Error pattern: 3 lines only -- What / Why: reason / Fix: /mos:command
+
+## Pre-flight: Room Check
+
+Resolve the active room using the shared helper:
+
+```bash
+bash ${MINDRIAN_OS_ROOT:-${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/resolve-room
+```
+
+If no Data Room is found, show the 3-line error and STOP:
+
+    x No Data Room found
+      Why: .mindrian/ missing in the current or parent directories
+      Fix: /mos:new-project
+
+If the room exists but `.mindrian/whitespace-embeddings.json` is absent, show the 3-line error and STOP:
+
+    x No whitespace embeddings
+      Why: Diagnostics requires an embedded artifact corpus at .mindrian/whitespace-embeddings.json
+      Fix: /mos:whitespace map
+
+## Step 1: Run Pipeline
+
+Invoke the dispatcher:
+
+```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/diagnostics-command.cjs ROOM_DIR
+```
+
+The dispatcher runs four Python scripts in sequence:
+
+- `compute-disruption-index.py` (Funk and Owen-Smith 2017 CD index)
+- `compute-blindspot-mass.py` (Good-Turing coverage estimation)
+- `compute-element-novelty.py` (centroid-distance novelty)
+- `compute-bayesian-surprise.py` (leave-one-out cosine shift)
+
+Each emits a JSON file in `.mindrian/`. The dispatcher reads the metadata scalars and composes the Shape E report.
+
+## Step 2: Render 4-Zone Output
+
+### Zone 1: Header Panel
+
+One line naming the room and the panel:
+
+    -- [Room Name] -- Diagnostics -- Wave-1 Algorithmic Fingerprint --
+
+### Zone 2: Content Body (Shape E Action Report)
+
+A 4-row metric table. The dispatcher produces the table verbatim; pass it through.
+
+Example from mindrianOS room (2026-04-23 smoke test):
+
+```
+  Metric                Value         Interpretation
+  ------------------------------------------------------------------------
+  Disruption (CD)       -0.7092, room is consolidating, not disrupting
+  Coverage              0.667, partial coverage, 33 pct uncovered
+  Element Novelty       0.083 mean, low diversity, artifacts cluster tightly
+  Bayesian Surprise     max 0.312, moderate surprise in recent additions
+```
+
+### Zone 3: Intelligence Strip (conditional, max 3 signals)
+
+Threshold rules evaluated by the dispatcher:
+
+- `warning` CD below -0.3 -- Strong consolidation signal
+- `empty-square` Coverage below 0.4 -- Low coverage, X pct of problem space unmapped
+- `warning` Novelty mean below 0.1 -- Low diversity, artifacts cluster tightly
+- `lightning` Surprise max above 0.5 -- Recent additions carry high information gain
+
+Only triggered signals render. A healthy room may show zero signals.
+
+### Zone 4: Action Footer (never omit)
+
+Three arrow-prefixed follow-ups:
+
+    -> /mos:whitespace map                Inspect zones behind these scalars
+    -> /mos:whitespace discover           Run full Discovery Cycle
+    -> /mos:find-bottlenecks              Drill into reverse salients
+
+## Interpretation Guide
+
+| Metric | Range | What It Means |
+|--------|-------|---------------|
+| Disruption (CD) | -1.0 (pure consolidation) to +1.0 (pure disruption) | Funk and Owen-Smith 2017. Negative means the room's new ideas build on prior art (consolidating); positive means they displace prior art (disrupting). |
+| Coverage | 0.0 (nothing mapped) to 1.0 (fully mapped) | Good-Turing estimator. Fraction of the problem space covered by at least one artifact. 1 - coverage = fraction of the space that is unmapped blindspot. |
+| Element Novelty | 0.0 (identical clones) to 1.0 (maximally diverse) | Mean centroid-distance across artifacts. Low mean -> artifacts cluster tightly; high mean -> diverse perspectives. |
+| Bayesian Surprise | 0.0 (confirms prior) to ~1.0 (total break) | Leave-one-out cosine shift. Max across artifacts indicates the most information-dense recent addition. |
+
+## Error Handling
+
+Three-line pattern. Specific failures:
+
+1. Python not installed:
+
+        x Python runtime not found
+          Why: python3 not on PATH, cannot run Wave-1 scripts
+          Fix: install Python 3.10+ and re-run /mos:diagnostics
+
+2. No whitespace embeddings (pre-flight already catches this):
+
+        x No whitespace embeddings
+          Why: Diagnostics requires an embedded artifact corpus at .mindrian/whitespace-embeddings.json
+          Fix: /mos:whitespace map
+
+3. Partial results (one or more scripts failed but the dispatcher continued):
+
+    The dispatcher prints a yellow `Partial results:` block listing each metric that failed, with the script error message or the missing-field diagnostic. The rest of the Shape E report renders normally; the failed row shows `no signal` instead of a scalar.
+
+## Cross-Surface Adaptation
+
+- **CLI** -- Full power. Scripts invoked via Bash. 80-column terminal output. The Shape E table is rendered with the 12-glyph vocabulary.
+- **Desktop** -- Larry narrates conversationally in prose. Example: "Very simply, the room is consolidating (CD = -0.71), two-thirds mapped (coverage 0.667), low diversity (novelty 0.083), moderate recent surprise (max 0.312)." Numbers shown as markdown bullets, no tree chars per ui-system Section 9.
+- **Cowork** -- Same as CLI. The four metrics are shared via `00_Context/` for team visibility. The intelligence strip becomes a shared "room health" snapshot.
+
+## Canon References
+
+- **Canon Part 2 Engine 1:** Act 1 intelligence surface. These four scalars complete the Wave-1 quantitative layer that feeds Engine 2 (BONO Orchestration). Whitespace Map + Reverse Salient + Cross-Domain Match sit alongside them in Engine 1.
+- **Canon Part 7 Reuse Before Build:** /mos:diagnostics is net-new surface because the four Wave-1 scalars answer a question orthogonal to /mos:grade (qualitative stage rubric) and /mos:status (pipeline progression). Folding four numerical diagnostics into either would dilute both surfaces.
+- **Canon Part 8 Graph Boundary:** All four algorithms operate LOCAL-only on the room's embedding corpus. No user data egress to Brain. Zero Brain query sites in scripts/diagnostics-command.cjs -- verified by inspection.

package/skills/dial-memory-refresh/SKILL.md

@@ -0,0 +1,116 @@
+---
+name: dial-memory-refresh
+description: Refresh the Dial Memory (auto) section in each room section's memory MD
+license: BSL-1.1. See LICENSE for complete terms (Business Source License 1.1, Change Date 2030-04-16 to Apache License 2.0).
+compatibility: Requires Claude Code (or a host implementing disable-model-invocation semantics); Tier-1 hook mechanics referenced in this skill.
+help_jtbd: "Refresh the human-readable Dial Memory section rendered from the graph."
+body_shape: E
+hitl_shape: "F.0"
+hitl_why: "It offers a single refresh action to approve or defer."
+# Phase 267.3-06, ruled in 267.3-CLASSIFICATION.md (Row 7): first delivery at commands/dial-memory-refresh.md:57, an F.0 Action Report of sections refreshed/skipped/failed, a regeneration-status report.
+interactive_first_reward: "--none (diagnostic surface)"
+argument-hint: "[--all | --section <slug>]"
+serves_jtbd: ["validate-idea", "audit-room"]
+teaching: "When a section's Dial Memory feels stale, /mos:dial-memory-refresh rebuilds the auto-section from the SELECTED_REACH / PIVOTED / DRSCH relationship layer. Human-authored prose stays byte-preserved."
+disable-model-invocation: false
+allowed-tools: Bash(node *), AskUserQuestion
+kind: utility
+frameworks: []
+produces: "room/*/FEYNMAN.md"
+inputs: []
+autonomous_safe: true
+# --- Phase 144.1 connector frontmatter ---
+connector:
+  connects_to_spine: true
+  sensor_triggers: []
+  reach_id: context_block
+  sub_mode: dial-refresh
+  framework: "Reverse Salient Analysis"
+  posture: hold
+  hierarchy_rank: 22
+  filing: memory_event_only
+  plan_gated: false
+  web_scope: null
+---
+
+<!-- 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:dial-memory-refresh
+
+Force-refresh the `## Dial Memory (auto)` sentinel-bounded section in each section's memory MD. The renderer reads the dial relationship layer (`SELECTED_REACH` / `PIVOTED` / `DRSCH` evidence edges) from `room.db` via the Phase 109 `navigation.cjs` chokepoint; the runner (a thin shim over the Phase 124 timeline-runner) writes back atomically with the human-authored body byte-preserved. This is the manual knob the user reaches for when the automatic session-start cascade is not enough.
+
+## Modes
+
+- `/mos:dial-memory-refresh` -- refresh every section in the active room (default; equivalent to `--all`).
+- `/mos:dial-memory-refresh --all` -- refresh every section in the active room.
+- `/mos:dial-memory-refresh --section <slug>` -- refresh one section.
+
+## Output
+
+F.0 Action Report (Canon Part 3 UI Ruling System; same shape as `/mos:feynman-timeline-refresh`). Three summary rows + a PER SECTION breakdown + a NEXT action footer:
+
+- Sections refreshed: N
+- Sections skipped: S
+- Sections failed: F
+
+## The 4 rendered components (MEMDIAL-02)
+
+The Dial Memory section is rendered FROM the graph (MEMDIAL-03; the graph is the source of truth) and carries all four components on every refresh:
+
+1. Available reaches -- the 6 canonical reach ids (the Phase 141 doctrine; Phase 148 D-09 raised 5 -> 6 with the hats reach).
+2. Last selected -- the most-recent `SELECTED_REACH` edge.
+3. Current recommended -- the frozen 0.70-gate result.
+4. Recent research conclusions -- surfaced from the `DRSCH` research-conclusion evidence edges; a clean placeholder when none exist.
+
+## Graceful degradation
+
+- **No active room**: exit 1 with "Set one with /mos:rooms switch <slug>".
+- **room.db missing**: single-line message, exit 0 (soft-fail; the dial-memory section is enrichment, not a gate).
+- **Runner exception per section**: the failure is logged via `dial_memory_refresh_failed` memory_event; the per-section row names the failure; other sections still process.
+
+## Canon references
+
+- **Canon Part 7 (Reuse Before Build)**: the runner is a THIN SHIM over `lib/core/feynman/timeline-runner.cjs` (Phase 124). No new atomic-write logic at the runner or command layer.
+- **Canon Part 8 (Graph Boundary)**: zero net-new Brain surface. The renderer reads ONLY `room.db` via `navigation.cjs`. The runner writes ONLY inside the dial sentinels.
+- **Canon Part 9 (Memory Locality)**: this is the Larry-explains face of the dial relationship layer -- the rendered section IS the human-readable view of the SQL graph.
+
+## Invocation
+
+```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/dial-memory-refresh-command.cjs $ARGUMENTS
+```
+
+## Examples
+
+```
+/mos:dial-memory-refresh
+/mos:dial-memory-refresh --all
+/mos:dial-memory-refresh --section problem-definition
+```
+
+## Exit codes
+
+| Exit | Meaning |
+| --- | --- |
+| 0 | Success (including soft-fail paths: room.db missing) |
+| 1 | Invocation error: no active room |
+
+## Cross-surface adaptation
+
+- **CLI**: full power. The slash command runs the dispatcher through the Bash tool. F.0 report renders in the terminal.
+- **Desktop**: the same slash command runs when the plugin is connected. Larry may narrate the result conversationally; the dispatcher is identical.
+- **Cowork**: same as CLI. Team members see the regenerated `## Dial Memory (auto)` block through the shared `00_Context/` snapshot on their next session-start.
+
+No surface-specific code exists anywhere in the dispatcher.