mindforge-cc 11.9.0 → 11.9.2

package/.claude/commands/mindforge/wf-verification-loop.md

@@ -0,0 +1,45 @@
+---
+description: "Implementer vs. scorer loop: revise each round, score against a fixed metric, keep re-attempting until the score is maxed or a round cap is hit"
+---
+# /mindforge:wf-verification-loop
+
+Runs the **Verification Loop** dynamic workflow.
+
+## Usage
+`/mindforge:wf-verification-loop [task description, or { task, metric, targetScore, maxRounds }]`
+
+## What it does
+- **Scope**: Pins the artifact, the metric, and the target score / max-round cap
+- **Round**: An implementer agent revises the artifact; a separate scorer agent grades it
+  against the same fixed rubric every round; repeats until the target score is hit, the
+  score plateaus for `plateauRounds` consecutive rounds, or `maxRounds` is reached
+- **Report**: Round-by-round score history plus the final artifact
+
+Unlike a fixed-count review pass, this loop's whole point is that the implementer keeps
+re-attempting the SAME task, judged by a separate scorer, until the score stops moving —
+not a one-shot audit.
+
+## Running
+
+Invoke via Claude Code's Workflow tool:
+
+```
+Workflow({
+  scriptPath: ".mindforge/dynamic-workflows/scripts/verification-loop.js",
+  args: {
+    task: "Tighten this system prompt for token efficiency without losing instruction coverage",
+    metric: "10 pts clarity, 10 pts token economy, 10 pts no lost instructions",
+    targetScore: 95,
+    maxRounds: 8
+  }
+})
+```
+
+`args` can also be a plain string task description if you want the scorer to infer and
+hold its own rubric. Defaults: `targetScore` 95, `maxRounds` 8, `plateauRounds` 2 (stop
+early if the score doesn't improve for 2 consecutive rounds).
+
+Or discover via CLI:
+```bash
+node bin/mindforge-cli.js workflow info verification-loop
+```

package/.claude/commands/mindforge/wf-workflow-optimizer.md

@@ -0,0 +1,51 @@
+---
+description: "Builder + scorer + a distinct process-optimizer role: proposes structural changes to the PROCESS itself across rounds, not just fixes to one output"
+---
+# /mindforge:wf-workflow-optimizer
+
+Runs the **Workflow Optimizer** dynamic workflow.
+
+## Usage
+`/mindforge:wf-workflow-optimizer [task family description, or { taskFamily, metric, maxRounds, priorHistory }]`
+
+## What it does
+- **Scope**: Pins the task family, the fixed outcome metric, and any prior round history
+  you supply (this script holds no state of its own between invocations)
+- **Cycle**: A builder agent runs under the *current* process instructions; a separate
+  scorer agent grades the output against the fixed metric; a **third, distinct
+  process-optimizer agent** — with no role in building or scoring — diagnoses *why* the
+  process produced that result and rewrites the builder's instructions for next round.
+  Repeats until the optimizer judges the process has converged or `maxRounds` is hit.
+- **Report**: The score trend across rounds alongside every round's structural diagnosis,
+  so you can see whether rising scores came from real process changes, not just luck.
+
+This is not "implementer vs. scorer, keep re-attempting the same task" (that's
+`/mindforge:wf-verification-loop`, one artifact, one metric, no separate optimizer role).
+This workflow's object of improvement is the **process itself** — the builder's own
+instructions evolve round to round, and the deliverable includes that evolution history,
+not just a final artifact.
+
+## Running
+
+Invoke via Claude Code's Workflow tool:
+
+```
+Workflow({
+  scriptPath: ".mindforge/dynamic-workflows/scripts/workflow-optimizer.js",
+  args: {
+    taskFamily: "Write a one-paragraph executive summary of a technical incident postmortem",
+    metric: "30 pts non-technical-reader clarity, 30 pts root-cause traceability, 40 pts action-item concreteness",
+    maxRounds: 5
+  }
+})
+```
+
+To continue optimizing the SAME process across separate sessions, pass the previous run's
+returned `processHistory` back in as `args.priorHistory` — the script itself has no
+filesystem access and keeps no state between invocations, so cross-session continuity is
+the caller's responsibility.
+
+Or discover via CLI:
+```bash
+node bin/mindforge-cli.js workflow info workflow-optimizer
+```

package/.claude/commands/mindforge/wf-writer-reviewer.md

@@ -0,0 +1,30 @@
+---
+description: "Anthropic Writer/Reviewer pattern: implement in Context A → fresh Context B reviews the diff"
+---
+# /mindforge:wf-writer-reviewer
+
+Runs the **Writer Reviewer** dynamic workflow.
+
+## Usage
+`/mindforge:wf-writer-reviewer <implementation task description>`
+
+## What it does
+- **Implement**: Writer agent implements the requested change
+- **Review**: Fresh reviewer agent inspects only the diff without implementation context
+- **Verdict**: Accept / request-changes verdict with specific actionable feedback
+
+## Running
+
+Invoke via Claude Code's Workflow tool:
+
+```
+Workflow({
+  scriptPath: ".mindforge/dynamic-workflows/scripts/writer-reviewer.js",
+  args: "<your input>"
+})
+```
+
+Or discover via CLI:
+```bash
+node bin/mindforge-cli.js workflow info writer-reviewer
+```

package/.mindforge/MINDFORGE-SCHEMA.json

@@ -1,33 +1,146 @@
 {
   "$schema": "http://json-schema.org/draft-07/schema#",
   "title": "MindForge Project Configuration Schema",
-  "description": "JSON Schema for MINDFORGE.md key-value settings",
+  "description": "JSON Schema for MINDFORGE.md bracketed parameter assignments ([KEY] = value)",
   "type": "object",
+  "required": [
+    "VERSION",
+    "REACTIVE_MODE",
+    "PLANNER",
+    "EXECUTOR",
+    "MIN_SOUL_SCORE"
+  ],
+  "recommended": [
+    "COST_WARN_USD",
+    "COST_HARD_LIMIT_USD",
+    "BLOCK_ON_SECURITY"
+  ],
   "properties": {
+    "NAME": { "type": "string", "description": "Project display name" },
+    "VERSION": {
+      "type": "string",
+      "pattern": "^\\d+\\.\\d+\\.\\d+$",
+      "description": "MindForge version this registry targets (must equal package.json version)"
+    },
+    "STABLE": { "type": "boolean", "description": "Registry is a released/stable configuration" },
+    "MODE": { "type": "string", "description": "Operating mode label (free-form)" },
+    "REQUIRED_CORE_VERSION": {
+      "type": "string",
+      "pattern": "^\\d+\\.\\d+\\.\\d+$",
+      "description": "Minimum MindForge core version required by this project"
+    },
+    "SOVEREIGN_IDENTITY": {
+      "type": "boolean",
+      "nonOverridable": true,
+      "description": "NON-OVERRIDABLE: identity verification is always required"
+    },
+    "SRE_LAYER_ENABLED": { "type": "boolean", "description": "Enable the SRE persona layer" },
+    "SWARM_ORCHESTRATION": { "type": "boolean", "description": "Enable dynamic swarm orchestration" },
+    "AUTO_SWARM_THRESHOLD": {
+      "type": "number", "minimum": 0, "maximum": 10,
+      "description": "Difficulty score above which a swarm is auto-spawned"
+    },
+    "CONTEXT7_DEPTH": {
+      "type": "enum", "values": ["MINIMAL", "STANDARD", "EXTENDED"],
+      "description": "Depth of Context7 knowledge injection"
+    },
+    "DYNAMISM_LEVEL": {
+      "type": "number", "minimum": 1, "maximum": 5,
+      "description": "How aggressively the engine reshapes its own plan (1=static, 5=fully dynamic)"
+    },
+    "REACTIVE_MODE": { "type": "boolean", "description": "Enable the reactive (auto-discovery) engine" },
+    "CONTEXT_SHARDING": { "type": "boolean", "description": "Enable the sharded memory loop" },
+    "ADS_LOOP_ENFORCED": { "type": "boolean", "description": "Enforce the adversarial decision loop before architectural change" },
+    "TEMPORAL_VISION": { "type": "boolean", "description": "Enable temporal hindsight / steering" },
+    "RAG_2_AUTO_SHADOW": { "type": "boolean", "description": "Run RAG-2 retrieval in shadow mode" },
+    "ENABLE_ART_TRACING": { "type": "boolean", "description": "Enable agent reasoning trace (ART) capture" },
+    "ENABLE_ZTAI": {
+      "type": "boolean",
+      "nonOverridable": true,
+      "description": "NON-OVERRIDABLE: zero-trust agent identity cannot be bypassed"
+    },
+    "ZTAI_KEY_TYPE": { "type": "string", "description": "Declared ZTAI key algorithm label" },
+    "NEXUS_TRACE_RETENTION_DAYS": {
+      "type": "number", "minimum": 1, "maximum": 365,
+      "description": "Days of NexusTracer trace retention"
+    },
+    "CADIA_CORE": { "type": "boolean", "description": "Enable the CADIA core layer" },
+    "PQAS_ENFORCED": {
+      "type": "boolean",
+      "description": "Post-quantum attestation enforcement (SIMULATED/inactive by default — see bin/governance/quantum-crypto.js)"
+    },
+    "PROACTIVE_HOMING": { "type": "boolean", "description": "Enable proactive homing / next-task discovery" },
+    "PLANNER": { "type": "string", "description": "Model id for the planning persona (canonical: PLANNER_MODEL)" },
+    "EXECUTOR": { "type": "string", "description": "Model id for execution agents (canonical: EXECUTOR_MODEL)" },
+    "REVIEWER": { "type": "string", "description": "Model id for the code reviewer (canonical: REVIEWER_MODEL)" },
+    "VERIFIER": { "type": "string", "description": "Model id for the verification persona (canonical: VERIFIER_MODEL)" },
+    "SECURITY": { "type": "string", "description": "Model id for security review (canonical: SECURITY_MODEL)" },
+    "DEBUG": { "type": "string", "description": "Model id for the debug specialist (canonical: DEBUG_MODEL)" },
+    "RESEARCH": { "type": "string", "description": "Model id for the research agent (canonical: RESEARCH_MODEL)" },
+    "QA": { "type": "string", "description": "Model id for the QA engineer (canonical: QA_MODEL)" },
+    "QUICK": { "type": "string", "description": "Model id for tier-1 budget-bias tasks (canonical: QUICK_MODEL)" },
+    "COST_WARN_USD": {
+      "type": "number", "minimum": 0, "maximum": 10000,
+      "description": "Daily spend at which a cost warning is emitted"
+    },
+    "COST_HARD_LIMIT_USD": {
+      "type": "number", "minimum": 0, "maximum": 10000,
+      "description": "Daily hard cost limit in USD. DECLARED BUT NOT ENFORCED as of 11.9.2: bin/models/cost-tracker.js reads MODEL_COST_HARD_LIMIT_USD, which this registry does not declare, so preflight() returns early and no cap is applied. Do not treat this as a spend control. Tracked as COST-02."
+    },
+    "ADS_DEBATE_ROUNDS": {
+      "type": "number", "minimum": 1, "maximum": 10,
+      "description": "Red-team/blue-team debate rounds per architectural decision"
+    },
+    "MIN_SOUL_SCORE": {
+      "type": "number", "minimum": 0, "maximum": 10,
+      "nonOverridable": true,
+      "description": "NON-OVERRIDABLE: minimum SOUL score required for architectural changes"
+    },
+    "REQUIRE_ADR": { "type": "boolean", "description": "Require an ADR for architectural decisions" },
+    "BLOCK_ON_SECURITY": {
+      "type": "boolean",
+      "nonOverridable": true,
+      "description": "NON-OVERRIDABLE: security gate enforcement cannot be disabled"
+    },
+    "BROWSER_PORT": {
+      "type": "number", "minimum": 1024, "maximum": 65535,
+      "description": "Port for the browser/QA harness"
+    },
+    "DASHBOARD_PORT": {
+      "type": "number", "minimum": 1024, "maximum": 65535,
+      "description": "Port for the MindForge dashboard"
+    },
+    "API_URL": { "type": "string", "description": "Base URL of the project API under development" },
+    "SHARD_RETAIN_DAYS": {
+      "type": "number", "minimum": 1, "maximum": 365,
+      "description": "Days to retain context shards"
+    },
+    "FORBIDDEN": { "type": "string", "description": "Multi-line block of project prohibitions injected into every agent prompt" },
+    "INSTRUCTIONS": { "type": "string", "description": "Multi-line block of project instructions injected into every agent prompt" },
     "MINDFORGE_VERSION_REQUIRED": {
       "type": "string",
       "pattern": "^\\d+\\.\\d+\\.\\d+$",
       "description": "Minimum MindForge version required for this config"
     },
     "PLANNER_MODEL": {
-      "type": "enum",
-      "values": ["claude-opus-4-5", "claude-sonnet-4-5", "claude-haiku-4-5", "inherit"],
-      "description": "Claude model to use for the planning agent"
+      "type": "string",
+      "deprecated": true,
+      "description": "DEPRECATED long form of [PLANNER]. Model id for the planning agent (free-form: model ids churn faster than this schema)"
     },
     "EXECUTOR_MODEL": {
-      "type": "enum",
-      "values": ["claude-opus-4-5", "claude-sonnet-4-5", "claude-haiku-4-5", "inherit"],
-      "description": "Claude model to use for execution agents"
+      "type": "string",
+      "deprecated": true,
+      "description": "DEPRECATED long form of [EXECUTOR]. Model id for execution agents"
     },
     "REVIEWER_MODEL": {
-      "type": "enum",
-      "values": ["claude-opus-4-5", "claude-sonnet-4-5", "claude-haiku-4-5", "inherit"],
-      "description": "Claude model to use for the code reviewer"
+      "type": "string",
+      "deprecated": true,
+      "description": "DEPRECATED long form of [REVIEWER]. Model id for the code reviewer"
     },
     "SECURITY_MODEL": {
-      "type": "enum",
-      "values": ["claude-opus-4-5", "claude-sonnet-4-5", "claude-haiku-4-5", "inherit"],
-      "description": "Claude model to use for security review (recommend Opus for thoroughness)"
+      "type": "string",
+      "deprecated": true,
+      "description": "DEPRECATED long form of [SECURITY]. Model id for security review"
     },
     "TIER1_AUTO_APPROVE": {
       "type": "boolean",

package/.mindforge/config.json

@@ -1,11 +1,11 @@
 {
-  "version": "11.9.0",
+  "version": "11.9.2",
   "environment": "development",
   "governance": {
     "drift_threshold": 0.75,
     "critical_drift_threshold": 0.5,
     "res_threshold": 0.8,
-    "active_did": "did:mindforge:7672fba0-0c90-429f-8f22-2c6aa7d2bc63"
+    "active_did": "did:mindforge:766e07ac-8f56-4840-be61-316b90bbed35"
   },
   "revops": {
     "market_registry": {
@@ -125,7 +125,7 @@
       "project_weekly_warn_usd": 50,
       "project_weekly_hard_limit_usd": 200
     },
-    "ledger_path": ".mindforge/metrics/token-ledger.jsonl"
+    "ledger_path": ".mindforge/metrics/token-usage.jsonl"
   },
   "proactive_suggestions": {
     "enabled": true,

package/.mindforge/engine/cost-tracking/router.md

@@ -41,7 +41,7 @@ These always override the matrix:
 3. Map score to tier via decision matrix
 4. Apply override rules
 5. Check budget: if remaining budget < estimated cost, downgrade one tier with WARNING
-6. Log routing decision to token-ledger.jsonl
+6. Log routing decision to token-usage.jsonl
 7. Return selected model ID
 
 ## Budget Guard

package/.mindforge/engine/cost-tracking/token-ledger.md

@@ -5,43 +5,40 @@ Append-only ledger recording all token usage for analytics, budgeting, and optim
 
 ## Storage
 
-- Location: `.mindforge/metrics/token-ledger.jsonl`
+- Location: `.mindforge/metrics/token-usage.jsonl` (canonical shape: `bin/models/usage-record.js`)
 - Format: JSON Lines (one entry per model interaction)
 - Rotation: Archive entries older than 30 days to `.mindforge/metrics/archive/`
 - Retention: Archives kept for 90 days, then deleted
 
 ## Entry Format
 
-Each line in the ledger is a complete JSON object:
+Each line in the ledger is a complete JSON object. The authoritative shape is
+`bin/models/usage-record.js`; this is what `bin/models/cost-tracker.js`
+actually appends today (all five providers in `bin/models/*-provider.js` emit it):
 
 ```json
 {
-  "id": "txn-[uuid]",
-  "timestamp": "2026-05-25T10:30:00Z",
-  "session_id": "session-abc123",
-  "task_id": "task-def456",
-  "phase": "execute",
   "model": "claude-sonnet-4-6",
-  "tier": "standard",
-  "routing_reason": "difficulty_score_5_multi_file",
-  "tokens": {
-    "input": 12500,
-    "output": 3200,
-    "cached_input": 8000,
-    "total": 15700
-  },
+  "input_tokens": 12500,
+  "output_tokens": 3200,
+  "cache_read_input_tokens": 8000,
+  "cache_creation_input_tokens": 0,
   "cost_usd": 0.085,
-  "budget_remaining": {
-    "session": 4.915,
-    "project_weekly": 49.915
-  },
-  "task_type": "implementation",
-  "files_touched": 3,
-  "skills_loaded": ["code-quality", "testing-standards"],
-  "outcome": "success"
+  "task_name": "Plan 3-04",
+  "session_id": "session-abc123",
+  "phase": 3,
+  "date": "2026-05-25",
+  "timestamp": "2026-05-25T10:30:00.000Z"
 }
 ```
 
+`cost_usd` is the ONLY cost field. `total_cost_usd` belongs to cross-review
+reports (`bin/review/cross-review-engine.js`) and must never appear here.
+
+Not yet emitted (aspirational — do not read these): `id`, `task_id`, `tier`,
+`routing_reason`, `budget_remaining`, `task_type`, `files_touched`,
+`skills_loaded`, `outcome`.
+
 ## Reporting Queries
 
 The `/mindforge:cost-report` command reads this ledger to produce:
@@ -71,7 +68,7 @@ debugging       | $0.85    | 32,000     | 8
 
 ## Integration
 
-- Written to by the budget-enforcer after every model interaction
+- Written to by `bin/models/cost-tracker.js` `record()` after every model interaction (called from `bin/models/model-client.js:77`)
 - Read by `/mindforge:cost-report` command
 - Summarized weekly into `.mindforge/metrics/weekly-cost-summary.json`
 - Referenced by AgRevOps dashboard for ROI tracking

package/.mindforge/memory/sync-manifest.json

@@ -1,6 +1,6 @@
 {
   "pattern-library.jsonl": {
-    "lastSync": "2026-07-28T09:19:09.269Z",
+    "lastSync": "2026-08-16T19:50:41.118Z",
     "localCount": 1
   }
 }

package/.mindforge/metrics/METRICS-SCHEMA.md

@@ -1,13 +1,22 @@
 # MindForge Metrics — Schema Reference
 
 ## Files
-- `session-quality.jsonl`
-- `phase-metrics.jsonl`
-- `skill-usage.jsonl`
-- `compaction-quality.jsonl`
+- `session-quality.jsonl` — **no producer in this repo**; every reference to it
+  (`bin/dashboard/metrics-aggregator.js`, `sdk/src/client.ts`,
+  `mcp-server/src/vendor/client.ts`) is a READ. Expected to be supplied by the
+  agent/host; MindForge only consumes it.
+- `phase-metrics.jsonl` — no producer in this repo
+- `skill-usage.jsonl` — no producer in this repo
+- `compaction-quality.jsonl` — no producer in this repo
+- `token-usage.jsonl` — **code-written** by `bin/models/cost-tracker.js`; canonical
+  shape is `bin/models/usage-record.js`. Cost lives in `cost_usd` and nowhere else.
+  Gitignored and excluded from the npm tarball.
 
 All files are append-only JSONL.
 
+Per-session cost is NOT a `session-quality.jsonl` field: the dashboard joins it
+from `token-usage.jsonl` on `session_id`.
+
 ## session-quality.jsonl fields
 - session_id, date, phase, developer_id
 - tasks_attempted, tasks_completed, tasks_failed

package/.mindforge/personas/cost-optimizer.md

@@ -39,7 +39,7 @@ Map score to model tier via the routing decision matrix.
 </step>
 
 <step name="check_budget">
-Read token-ledger.jsonl for current session/project spend.
+Read token-usage.jsonl for current session/project spend.
 Compare against budget limits in config.json.
 If approaching warn threshold: flag to user.
 </step>
@@ -66,6 +66,6 @@ Recommend routing adjustments for next session.
 - NEVER skip security overrides to save money (auth/payment always >= standard tier)
 - NEVER exceed hard budget limit without explicit user approval
 - NEVER silently downgrade model quality — always inform
-- Track every model interaction in token-ledger.jsonl
+- Track every model interaction in token-usage.jsonl
 - Report cost transparency in every session summary
 </critical_rules>

package/.mindforge/personas/doc-auditor.md

@@ -35,7 +35,7 @@ updates every time code changes, it needs automation or deletion.
 <process>
 <step name="inventory">
 Identify all documentation files in the project:
-- README.md, CONTRIBUTING.md, CHANGELOG.md
+- README.md, CONTRIBUTING.md, CHANGELOG.md (rolling window), changelogs/ (full archive + README.md index)
 - docs/ directory (all files)
 - Inline API documentation (JSDoc, docstrings)
 - Architecture decision records (ADRs)

package/.mindforge/personas/monorepo-architect.md

@@ -156,7 +156,8 @@ git commit -m "feat(api)!: change auth header" → major bump
 - Build all packages in dependency order
 - Publish to npm (private registry or public)
 - Tag git commit with version
-- Generate CHANGELOG.md
+- Generate `changelogs/vX.Y.Z.md` for the release, then prepend the same entry
+  (or a one-line summary + link) into root `CHANGELOG.md`'s rolling window
 </step>
 
 <step name="developer_experience">

package/.mindforge/personas/multi-model-bridge.md

@@ -82,5 +82,5 @@ Report to user with:
 - ALWAYS attribute responses to their source model
 - Maximum 2000 tokens per external prompt (cost control)
 - Maximum 3 consultations per session (rate limiting)
-- Log every external call in token-ledger.jsonl
+- Log every external call in token-usage.jsonl
 </critical_rules>

package/.mindforge/personas/release-manager.md

@@ -8,7 +8,7 @@ color: blue
 <role>
 You are the MindForge Release Manager. You own the transition from "Verification" to "Production."
 Your job is to ensure that MindForge versions are meaningful, stable, and perfectly documented.
-You are the final gatekeeper of the `CHANGELOG.md` and version tags.
+You are the final gatekeeper of the `CHANGELOG.md` rolling window, the `changelogs/` archive, and version tags.
 </role>
 
 <why_this_matters>
@@ -41,7 +41,9 @@ Check `STATE.md` to ensure all active workstreams are merged or paused.
 <step name="changelog_synthesis">
 Ingest the `SUMMARY.md` files from all completed phases.
 Group changes into: Added, Changed, Fixed, and Security.
-Write the entry to `CHANGELOG.md` using the Keep a Changelog standard.
+Create `changelogs/vX.Y.Z.md` for the new release using the Keep a Changelog standard,
+then prepend the same entry (or a one-line summary + link) to root `CHANGELOG.md`,
+trimming the oldest entry if the rolling window exceeds 10 versions.
 </step>
 
 <step name="version_bumping">
@@ -107,7 +109,7 @@ Draft the Pull Request (PR) or Release Note description.
 
 <success_criteria>
 - [ ] Version bump follows SemVer
-- [ ] CHANGELOG.md updated and logically grouped
+- [ ] changelogs/vX.Y.Z.md created and CHANGELOG.md rolling window updated (logically grouped)
 - [ ] All verification docs (UAT/Security) reviewed
 - [ ] Git tag created or staged
 - [ ] PR description finalized

package/.mindforge/skills/cost-aware-routing/SKILL.md

@@ -62,7 +62,7 @@ for cost-performance tradeoffs, or when budget limits are approaching.
 ### Budget Monitoring
 
 Check budget status regularly:
-- Session budget remaining: from token-ledger.jsonl
+- Session budget remaining: from token-usage.jsonl
 - Warning threshold: `[COST_WARN_USD]` from config
 - Hard limit: `[COST_HARD_LIMIT_USD]` from config
 
@@ -72,12 +72,12 @@ Check budget status regularly:
 - Never exceed hard limit without explicit user approval
 
 ### After any task
-- Log actual model used + tokens consumed to token-ledger.jsonl
+- Log actual model used + tokens consumed to token-usage.jsonl
 - Compare actual vs optimal tier (for future routing accuracy)
 - Report cost in session summary
 
 ## Self-check before task completion
 - [ ] Did I log the model routing decision with rationale?
-- [ ] Did I record actual token usage in token-ledger.jsonl?
+- [ ] Did I record actual token usage in token-usage.jsonl?
 - [ ] Did I check remaining budget against session/project limits?
 - [ ] Did I flag any tasks where a cheaper model could have been used?

package/.mindforge/skills/documentation/SKILL.md

@@ -75,7 +75,10 @@ export function verifyAccessToken(token: string): TokenPayload { ... }
 ```
 
 ## Changelog discipline
-Every user-visible change must appear in CHANGELOG.md before release.
+Every user-visible change must appear in the changelog before release: create
+`changelogs/vX.Y.Z.md` for the new release, then prepend the same entry (or a
+one-line summary + link) into root `CHANGELOG.md`'s rolling window, dropping
+the oldest entry if the window exceeds 10 versions.
 Format follows Keep a Changelog (keepachangelog.com).
 Categories: Added, Changed, Deprecated, Removed, Fixed, Security.
 

package/.mindforge/skills/multi-llm-consult/SKILL.md

@@ -57,7 +57,7 @@ multiple AI providers, or when the user explicitly requests cross-model consulta
    ```
 
 ### During consultation
-- Log every external call in token-ledger.jsonl (model, tokens, cost)
+- Log every external call in token-usage.jsonl (model, tokens, cost)
 - Never send more than 2000 tokens to external models per consultation
 - If a model is unavailable: skip it, note in output, continue with available models
 - Respect rate limits — max 3 consultations per session
@@ -70,6 +70,6 @@ multiple AI providers, or when the user explicitly requests cross-model consulta
 
 ## Self-check before task completion
 - [ ] Did I sanitize the prompt before sending to external models?
-- [ ] Did I log every external call in token-ledger.jsonl?
+- [ ] Did I log every external call in token-usage.jsonl?
 - [ ] Did I attribute responses to their source model (no unattributed blending)?
 - [ ] Did I remind the user that external opinions are advisory?