wogiflow 2.24.0 → 2.25.1

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (13) hide show
  1. package/.claude/commands/wogi-debug-hypothesis.md +1 -1
  2. package/.claude/commands/wogi-decide.md +40 -0
  3. package/.claude/commands/wogi-init.md +29 -0
  4. package/.claude/commands/wogi-learn.md +46 -0
  5. package/.claude/commands/wogi-onboard.md +26 -0
  6. package/.claude/commands/wogi-peer-review.md +5 -3
  7. package/.claude/commands/wogi-triage.md +57 -0
  8. package/package.json +2 -2
  9. package/scripts/flow-completion-truth-gate.js +130 -0
  10. package/scripts/flow-config-defaults.js +25 -0
  11. package/scripts/flow-extraction-review.js +18 -3
  12. package/scripts/flow-morning.js +12 -3
  13. package/scripts/flow-session-end.js +12 -3
package/.claude/commands/wogi-debug-hypothesis.md CHANGED
@@ -176,7 +176,7 @@ After all agents complete, display the consolidated results:
176
176
 
177
177
  ### Step 4: Hypothesis Adversary (v2.23.0+ — MANDATORY unless `--no-adversary`)
178
178
 
179
- After consolidation, spawn a single Agent (different `model` param if `config.hybrid.enabled`, else same) with this prompt:
179
+ After consolidation, spawn a single Agent on a DIFFERENT model (default `sonnet` via `config.researchReasoningGate.tier3.adversaryModel` — canonical cross-command adversary key, same as `/wogi-peer-review`, `/wogi-learn`, `/wogi-decide`) with this prompt:
180
180
 
181
181
  ```
182
182
  You are the hypothesis adversary.
package/.claude/commands/wogi-decide.md CHANGED
@@ -304,6 +304,46 @@ In `config.json`:
304
304
  }
305
305
  ```
306
306
 
307
+ ## Rule-Creation Adversary (v2.25.0+ — OPTIONAL but recommended for ambiguous rules)
308
+
309
+ When creating a non-trivial rule (anything beyond pure preference-setting like "always use semicolons"), spawn an adversary on a different model (default `sonnet` via `config.researchReasoningGate.tier3.adversaryModel`) to stress-test the proposed rule BEFORE it lands in `decisions.md`.
310
+
311
+ ```
312
+ Spawn Agent (subagent_type: general-purpose, model: <adversaryModel>):
313
+
314
+ Input:
315
+ Proposed rule title: <title>
316
+ Proposed rule body: <body>
317
+ User's original phrasing: <literal request>
318
+
319
+ Prompt:
320
+ You are the rule-creation adversary.
321
+ 1. Edge cases: name 3 situations where following this rule would produce
322
+ worse outcomes than NOT following it.
323
+ 2. Interpretation: are there 2+ reasonable interpretations? If yes, list
324
+ them and pick the one the user most likely meant.
325
+ 3. Scope creep: could this rule be over-applied to situations the user
326
+ didn't intend? Suggest scope qualifiers.
327
+ 4. Verdict:
328
+ - ACCEPT — ship as-is
329
+ - CLARIFY — multiple interpretations; ask user
330
+ - NARROW — over-application risk; add scope qualifiers
331
+ - REJECT — edge cases dominate; more harm than good
332
+
333
+ Output JSON: {
334
+ "verdict", "edge_cases", "interpretations",
335
+ "scope_qualifiers", "suggested_revision"
336
+ }
337
+ ```
338
+
339
+ Process:
340
+ - **ACCEPT** → proceed with rule creation
341
+ - **CLARIFY** → ask user to pick interpretation
342
+ - **NARROW** → show scope qualifier; ask user to approve
343
+ - **REJECT** → surface edge cases; require explicit override
344
+
345
+ Fail-open: adversary unavailable → proceed with standard flow. User confirmation is still present.
346
+
307
347
  ## Files
308
348
 
309
349
  | Action | File |
package/.claude/commands/wogi-init.md CHANGED
@@ -1209,3 +1209,32 @@ Say "show me the rules" or "what patterns are we using?" anytime.
1209
1209
  ### If user cancels mid-wizard
1210
1210
  - Save progress to `.workflow/state/setup-progress.json`
1211
1211
  - Next run can offer to resume
1212
+
1213
+ ## v2.25.0+ — Modern Config Scaffolding (MANDATORY)
1214
+
1215
+ New projects MUST be initialized with the following modern-stack config blocks explicitly written to `.workflow/config.json` so users can see + tune them (defaults-only inheritance is fine for behavior, but visibility matters for learning):
1216
+
1217
+ ```json
1218
+ {
1219
+ "intentGroundedReasoning": { "enabled": true },
1220
+ "taskBoundaryReset": {
1221
+ "enabled": true,
1222
+ "maxRestartsPerSession": 50
1223
+ },
1224
+ "storyFlow": {
1225
+ "consumerImpactAnalysis": { "enabled": true, "breakingThreshold": 5 },
1226
+ "scopeConfidenceAudit": { "enabled": true },
1227
+ "itemReconciliation": { "enabled": true, "minItems": 3 }
1228
+ },
1229
+ "longInputGate": { "enabled": true, "lineThreshold": 40 },
1230
+ "researchReasoningGate": {
1231
+ "enabled": true,
1232
+ "tier2": { "enabled": true },
1233
+ "tier3": { "enabled": true, "adversaryModel": "sonnet" }
1234
+ }
1235
+ }
1236
+ ```
1237
+
1238
+ These capabilities (IGR, task-boundary restart, P0 story gates, long-input routing, research reasoning gate) have proven out in 2.22+ releases. New users should NOT have to manually enable them via `flow migrate-igr` or equivalent — they are active from the first session.
1239
+
1240
+ If onboarding a workspace (multi-repo), also ensure `workspace.autoPickupChannelDispatches: true` and the 2.22.x restart-handoff settings are present.
package/.claude/commands/wogi-learn.md CHANGED
@@ -275,6 +275,52 @@ In `config.json`:
275
275
  }
276
276
  ```
277
277
 
278
+ ## Promotion Adversary (v2.25.0+ — MANDATORY)
279
+
280
+ Before promoting a pattern from `feedback-patterns.md` to `decisions.md`, run a **Promotion Adversary** on a different model. Rationale: same-model self-critique rubber-stamps. The adversary checks whether the N events that triggered promotion share an actual root cause (genuine recurrence) vs. superficial similarity with different underlying causes (false recurrence — common when the pattern detector just matched keywords).
281
+
282
+ ```
283
+ Spawn Agent (subagent_type: general-purpose,
284
+ model: config.researchReasoningGate.tier3.adversaryModel, default 'sonnet'):
285
+
286
+ Input:
287
+ Proposed rule: <title + body>
288
+ Triggering events: [
289
+ { date, request, correction },
290
+ { date, request, correction },
291
+ { date, request, correction }
292
+ ]
293
+
294
+ Prompt:
295
+ You are the rule-promotion adversary.
296
+ Do these N events actually share a root cause, or are they superficially
297
+ similar events with different underlying issues?
298
+
299
+ 1. For each event: describe the root cause in your own words.
300
+ 2. List what's common to all N root causes.
301
+ 3. List what's different between them.
302
+ 4. Verdict:
303
+ - SAME_PATTERN — genuine recurrence; rule is well-founded
304
+ - MIXED — N-1 match but one event has a different root cause
305
+ - DIFFERENT — surface-similar only; no unifying pattern
306
+
307
+ Output JSON:
308
+ {
309
+ "verdict": "SAME_PATTERN" | "MIXED" | "DIFFERENT",
310
+ "root_causes": [...],
311
+ "commonalities": [...],
312
+ "differences": [...],
313
+ "suggested_rule_scope": "as_proposed" | "narrower" | "split_into_N"
314
+ }
315
+ ```
316
+
317
+ Process the verdict:
318
+ - **SAME_PATTERN** → proceed with promotion as-is
319
+ - **MIXED** → ask the user: "Adversary flags event #X as different root cause. Promote rule anyway, narrow scope, or split into multiple rules?"
320
+ - **DIFFERENT** → DO NOT auto-promote. Surface adversary output; require explicit user confirmation.
321
+
322
+ Fail-open: if adversary cannot be spawned (missing API key, network), proceed with standard promotion and log a warning. The threshold check + user confirmation still apply.
323
+
278
324
  ## Files
279
325
 
280
326
  | Action | File |
package/.claude/commands/wogi-onboard.md CHANGED
@@ -1077,3 +1077,29 @@ AskUserQuestion({
1077
1077
  }]
1078
1078
  });
1079
1079
  ```
1080
+
1081
+ ## v2.25.0+ — Modern Config Scaffolding (MANDATORY)
1082
+
1083
+ When generating `.workflow/config.json` for a fresh project, include these 2.22+ capability blocks so new users inherit the current-best defaults:
1084
+
1085
+ ```json
1086
+ {
1087
+ "intentGroundedReasoning": { "enabled": true },
1088
+ "taskBoundaryReset": { "enabled": true, "maxRestartsPerSession": 50 },
1089
+ "storyFlow": {
1090
+ "consumerImpactAnalysis": { "enabled": true, "breakingThreshold": 5 },
1091
+ "scopeConfidenceAudit": { "enabled": true },
1092
+ "itemReconciliation": { "enabled": true, "minItems": 3 }
1093
+ },
1094
+ "longInputGate": { "enabled": true, "lineThreshold": 40 },
1095
+ "researchReasoningGate": {
1096
+ "enabled": true,
1097
+ "tier2": { "enabled": true },
1098
+ "tier3": { "enabled": true, "adversaryModel": "sonnet" }
1099
+ }
1100
+ }
1101
+ ```
1102
+
1103
+ These drive IGR (Architect + Adversary + Truth Gate), task-boundary context reset via the `wogi-claude` wrapper, `/wogi-story` P0 spec gates, auto-routing of long inputs to `/wogi-extract-review`, and the research reasoning gate's assumption-surfacing + cross-model adversary. All have proven out across the 2.22.x release series; new users should not have to discover them one at a time.
1104
+
1105
+ For multi-repo workspaces, also scaffold `workspace.autoPickupChannelDispatches: true` and leave the other `workspace.*` defaults intact — they include the 2.22.2 restart-handoff protocol.
package/.claude/commands/wogi-peer-review.md CHANGED
@@ -47,9 +47,11 @@ Models are selected once per session and remembered for subsequent runs.
47
47
  ├─────────────────────────────────────────────────────────┤
48
48
  │ 1. Collect code changes (git diff or specified files) │
49
49
  │ 2. Classify change size → effort tier: │
50
- │ L0/L1 (>10 files) → opus-4-7 xhigh │
50
+ │ L0/L1 (>10 files) → opus (latest) xhigh │
51
51
  │ L2 (3-10 files) → sonnet medium │
52
52
  │ L3 (<3 files) → haiku medium │
53
+ │ (Model IDs resolve from config.models — avoid │
54
+ │ hardcoding model version in this doc.) │
53
55
  │ 3. Generate improvement-focused prompt │
54
56
  │ 4. If includeClaude enabled: │
55
57
  │ - Launch Claude review (Task agent, Explore type) │
@@ -96,7 +98,7 @@ analysis, EACH carrying an explicit evidence tier.
96
98
 
97
99
  ## Synthesis Adversary (v2.23.0+ — MANDATORY unless `--no-adversary`)
98
100
 
99
- After initial synthesis, spawn a single adversary agent on a DIFFERENT model from the synthesizer (default: if synthesizer is Opus, adversary is Sonnet; config via `peerReview.adversaryModel`). Prompt:
101
+ After initial synthesis, spawn a single adversary agent on a DIFFERENT model from the synthesizer (default `sonnet`; override via the canonical `config.researchReasoningGate.tier3.adversaryModel` — same key used by `/wogi-debug-hypothesis`, `/wogi-learn`, `/wogi-decide`). Prompt:
100
102
 
101
103
  ```
102
104
  You are the synthesis adversary.
@@ -199,7 +201,7 @@ For manual review (no API keys needed): `/wogi-peer-review --manual`
199
201
  | `--verbose` | Show full model responses |
200
202
  | `--create-tasks` | Auto-create tasks for strong agreements |
201
203
  | `--no-adversary` | Skip the v2.23.0 synthesis adversary (not recommended for L0/L1 diffs) |
202
- | `--adversary-model <id>` | Override adversary model (default: cross-model from synthesizer) |
204
+ | `--adversary-model <id>` | Override adversary model (default: `config.researchReasoningGate.tier3.adversaryModel`, usually `sonnet`) |
203
205
  | `--effort <level>` | Override effort tier (low/medium/high/xhigh/max) — otherwise derived from diff size |
204
206
 
205
207
  ARGUMENTS: {args}
package/.claude/commands/wogi-triage.md CHANGED
@@ -356,3 +356,60 @@ Each finding is displayed using these fields from `last-review.json`:
356
356
  | File | `finding.file` + `finding.line` | "src/api.ts:45" |
357
357
  | Issue | `finding.issue` | "Raw JSON.parse without try-catch" |
358
358
  | Recommendation | `finding.recommendation` | "Use safeJsonParse from flow-utils.js" |
359
+
360
+ ## Anti-Deferral Enforcement (v2.25.0+ — two layers)
361
+
362
+ The **Review-Findings Anti-Deferral Rule** (`.workflow/state/decisions.md`, 2026-04-15) gets two complementary enforcement layers. One mechanical (an actual gate in the codebase), one AI-followed (a protocol documented here that the triage flow honors).
363
+
364
+ ### Layer 1 — Mechanical gate (v2.25.1+)
365
+
366
+ `scripts/flow-completion-truth-gate.js` exports `parseCommitMessageClaims()` and `verifyCommitMessageAgainstDiff()`. Callers pass a commit message and the staged diff (or changed-files list); the function parses finding IDs (`F1`/`M1`/`SEC-001`), task IDs (`wf-XXXXXXXX` after fix/close/resolve verbs), and file-path mentions, then checks each against the diff. Any unverified claim surfaces as a blocking prompt with three remediation options. This is real code, callable from pre-commit hooks, `flow-done.js`, or the triage flow itself.
367
+
368
+ Example usage:
369
+ ```javascript
370
+ const { verifyCommitMessageAgainstDiff, formatMissingClaimsMessage } =
371
+ require('wogiflow/scripts/flow-completion-truth-gate');
372
+
373
+ const result = verifyCommitMessageAgainstDiff(commitMsg, { diffText, changedFiles });
374
+ if (!result.ok) {
375
+ console.error(formatMissingClaimsMessage(result));
376
+ // Block + remediate
377
+ }
378
+ ```
379
+
380
+ ### Layer 2 — AI-followed protocol (documentation)
381
+
382
+ The rest of the triage flow is a protocol the AI follows. It is NOT automatically enforced by a hook — the historical v2.17.4 incident showed that doc-only protocols can be violated. The mechanical gate above closes the most damaging failure mode (commit message / diff mismatch). The AI-followed rules below cover the earlier stages:
383
+
384
+ 1. **Defer requires explicit user confirmation + reason.** The triage flow prompts when proposing to defer:
385
+ ```
386
+ Defer finding wf-review-XXXX?
387
+ Severity: HIGH
388
+ Reason required: [user input]
389
+ [Confirm defer] [Cancel — fix now]
390
+ ```
391
+ Auto-defer without reason is forbidden by this protocol.
392
+
393
+ 2. **"Fix all" / "Option 1" means fix ALL.** If the user requests bulk processing:
394
+ - Ship a fix for every finding with evidence-tier ≥ 1
395
+ - If any finding is too large, STOP and ask: "Finding X requires ~Y minutes of work. Ship now, split to its own release, or defer (needs reason)?"
396
+ - Never silently convert a finding to "deferred" in commit messages or release notes
397
+
398
+ 3. **Triage output includes a Deferral Audit Trail**:
399
+ ```
400
+ ━━━ TRIAGE SUMMARY ━━━
401
+ Fixed: 12
402
+ Deferred (with reasons): 2
403
+ • M3 — "requires restructure, tracked as wf-XXXXXXXX" (user-confirmed)
404
+ • L5 — "out of scope for current release" (user-confirmed)
405
+ Silently dropped: 0 ← MUST be 0
406
+ ━━━━━━━━━━━━━━━━━━━━━━
407
+ ```
408
+
409
+ ### Honest tradeoff
410
+
411
+ Layer 1 is genuinely mechanical — impossible for an AI to bypass without explicitly disabling the gate. Layer 2 is a protocol the AI can fail to follow if prompted poorly, distracted, or confused about priorities. Both matter; calling the whole system "architecturally impossible to bypass" would be inaccurate. The mechanical gate at least ensures that WHEN the AI writes a commit message, claimed fixes must actually appear in the diff.
412
+
413
+ Historical incident (v2.17.4 release, 2026-04-15): commit claimed "fix all findings" but M1 and M3 were silently dropped. Layer 1 would have caught that — the commit message mentioned M1 + M3 but the diff didn't. Layer 2 is the human-protocol reinforcement.
414
+
415
+ Skip via `config.triage.antiDeferralEnforcement.enabled: false` — note that this is currently a surface flag only (read by AI-followed protocol, not by the Layer 1 gate); to disable Layer 1 set `config.commitClaimsGate.enabled: false`.
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "wogiflow",
3
- "version": "2.24.0",
3
+ "version": "2.25.1",
4
4
  "description": "AI-powered development workflow management system with multi-model support",
5
5
  "main": "lib/index.js",
6
6
  "bin": {
@@ -10,7 +10,7 @@
10
10
  },
11
11
  "scripts": {
12
12
  "flow": "./scripts/flow",
13
- "test": "NODE_ENV=test node --test tests/auto-compact-prompt.test.js tests/flow-paths.test.js tests/flow-io.test.js tests/flow-config-loader.test.js tests/flow-damage-control.test.js tests/flow-output.test.js tests/flow-constants.test.js tests/flow-session-state.test.js tests/flow-hooks-integration.test.js tests/flow-utils.test.js tests/flow-security.test.js tests/flow-memory-db.test.js tests/flow-durable-session.test.js tests/flow-skill-matcher.test.js tests/flow-bridge.test.js tests/flow-proactive-compact.test.js tests/flow-cascade-completion.test.js tests/flow-capture-gate.test.js tests/flow-correction-detector-hybrid.test.js tests/flow-promote.test.js tests/flow-archive-runs.test.js tests/flow-memory.test.js tests/flow-hooks-pre-tool-helpers.test.js tests/flow-hooks-bugfix-scope-gate.test.js tests/flow-hooks-routing-gate.test.js tests/flow-hooks-phase-read-gate.test.js tests/flow-hooks-commit-log-gate.test.js tests/flow-hooks-deploy-gate.test.js tests/flow-hooks-todowrite-gate.test.js tests/flow-hooks-git-safety-gate.test.js tests/flow-hooks-scope-mutation-gate.test.js tests/flow-hooks-strike-gate.test.js tests/flow-hooks-component-check.test.js tests/flow-hooks-scope-gate.test.js tests/flow-hooks-implementation-gate.test.js tests/flow-hooks-research-gate.test.js tests/flow-hooks-loop-check.test.js tests/flow-hooks-manager-boundary-gate.test.js tests/flow-hooks-phase-gate.test.js tests/flow-hooks-pre-tool-orchestrator.test.js tests/flow-hooks-observation-capture.test.js tests/flow-hooks-task-gate.test.js tests/flow-durable-session-suspension.test.js tests/flow-health-mcp-scopes.test.js tests/flow-lean-config.test.js tests/flow-workspace-autopickup.test.js tests/flow-worker-boundary-gate.test.js tests/flow-worker-question-classifier.test.js tests/flow-completion-truth-gate-contradictions.test.js tests/flow-structure-sensor.test.js tests/flow-workspace-dispatch-tracking.test.js tests/flow-story-gates.test.js tests/flow-workspace-restart-handoff.test.js tests/flow-wogi-claude-wrapper.test.js tests/flow-wave1-integrations.test.js tests/flow-wave2-integrations.test.js && NODE_ENV=test node tests/run-quality-gates.test.js",
13
+ "test": "NODE_ENV=test node --test tests/auto-compact-prompt.test.js tests/flow-paths.test.js tests/flow-io.test.js tests/flow-config-loader.test.js tests/flow-damage-control.test.js tests/flow-output.test.js tests/flow-constants.test.js tests/flow-session-state.test.js tests/flow-hooks-integration.test.js tests/flow-utils.test.js tests/flow-security.test.js tests/flow-memory-db.test.js tests/flow-durable-session.test.js tests/flow-skill-matcher.test.js tests/flow-bridge.test.js tests/flow-proactive-compact.test.js tests/flow-cascade-completion.test.js tests/flow-capture-gate.test.js tests/flow-correction-detector-hybrid.test.js tests/flow-promote.test.js tests/flow-archive-runs.test.js tests/flow-memory.test.js tests/flow-hooks-pre-tool-helpers.test.js tests/flow-hooks-bugfix-scope-gate.test.js tests/flow-hooks-routing-gate.test.js tests/flow-hooks-phase-read-gate.test.js tests/flow-hooks-commit-log-gate.test.js tests/flow-hooks-deploy-gate.test.js tests/flow-hooks-todowrite-gate.test.js tests/flow-hooks-git-safety-gate.test.js tests/flow-hooks-scope-mutation-gate.test.js tests/flow-hooks-strike-gate.test.js tests/flow-hooks-component-check.test.js tests/flow-hooks-scope-gate.test.js tests/flow-hooks-implementation-gate.test.js tests/flow-hooks-research-gate.test.js tests/flow-hooks-loop-check.test.js tests/flow-hooks-manager-boundary-gate.test.js tests/flow-hooks-phase-gate.test.js tests/flow-hooks-pre-tool-orchestrator.test.js tests/flow-hooks-observation-capture.test.js tests/flow-hooks-task-gate.test.js tests/flow-durable-session-suspension.test.js tests/flow-health-mcp-scopes.test.js tests/flow-lean-config.test.js tests/flow-workspace-autopickup.test.js tests/flow-worker-boundary-gate.test.js tests/flow-worker-question-classifier.test.js tests/flow-completion-truth-gate-contradictions.test.js tests/flow-structure-sensor.test.js tests/flow-workspace-dispatch-tracking.test.js tests/flow-story-gates.test.js tests/flow-workspace-restart-handoff.test.js tests/flow-wogi-claude-wrapper.test.js tests/flow-wave1-integrations.test.js tests/flow-wave2-integrations.test.js tests/flow-wave3-integrations.test.js tests/flow-commit-claims-gate.test.js && NODE_ENV=test node tests/run-quality-gates.test.js",
14
14
  "test:syntax": "find scripts/ lib/ -name '*.js' -not -path '*/node_modules/*' -exec node --check {} +",
15
15
  "lint": "eslint scripts/ lib/ tests/",
16
16
  "lint:ci": "eslint scripts/ lib/ tests/ --max-warnings 0",
package/scripts/flow-completion-truth-gate.js CHANGED
@@ -614,6 +614,133 @@ function collectArrayEntries(obj, keys) {
614
614
  return out;
615
615
  }
616
616
 
617
+ // ============================================================
618
+ // Commit-vs-diff consistency scanner (v2.25.1 — H2b from Waves 1-3 review)
619
+ // ============================================================
620
+
621
+ /**
622
+ * Parse a commit message for "fixes X" / "closes X" / "F1, F2, M1" style claims
623
+ * that should be verifiable against the diff.
624
+ *
625
+ * Heuristics — conservative to avoid false positives:
626
+ * 1. Bracketed finding IDs: `F1`, `F2`, `M1`, `H3`, `L5`, or `SEC-001`/`PERF-002`
627
+ * 2. Task IDs: `wf-XXXXXXXX` that appear as "fixes wf-...", "closes wf-...", etc.
628
+ * 3. File paths mentioned in fix-context: "fixes `path/to/file.js`"
629
+ *
630
+ * Returns the structured claims a diff-consistency check can verify.
631
+ *
632
+ * @param {string} commitMessage
633
+ * @returns {{claims: Array<{kind: 'finding-id'|'task-id'|'file', value: string, raw: string}>}}
634
+ */
635
+ function parseCommitMessageClaims(commitMessage) {
636
+ const claims = [];
637
+ if (typeof commitMessage !== 'string' || commitMessage.trim().length === 0) {
638
+ return { claims };
639
+ }
640
+
641
+ // Finding IDs: F1, F2, M1, H3, L5, SEC-001, PERF-002, etc.
642
+ // - Single-letter + digits: match on word boundary
643
+ // - ALLCAPS-dashnum: SEC-001, PERF-002
644
+ const findingRe = /\b(?:F\d+|H\d+|M\d+|L\d+|[A-Z]{2,6}-\d+)\b/g;
645
+ for (const m of commitMessage.matchAll(findingRe)) {
646
+ claims.push({ kind: 'finding-id', value: m[0], raw: m[0] });
647
+ }
648
+
649
+ // Task IDs (wf-XXXXXXXX) — only count if preceded by fix/close/resolve verb
650
+ const taskRe = /\b(?:fix(?:es|ed)?|clos(?:es|ed)?|resolv(?:es|ed)?|address(?:es|ed)?)\s+(wf-[0-9a-f]{8})\b/gi;
651
+ for (const m of commitMessage.matchAll(taskRe)) {
652
+ claims.push({ kind: 'task-id', value: m[1], raw: m[0] });
653
+ }
654
+
655
+ // File paths in backticks after fix/address verbs: `fixes \`path/to/file.js\``
656
+ const fileRe = /(?:fix(?:es|ed)?|address(?:es|ed)?|updat(?:es|ed)?)\s+`([^`\n]{3,120})`/gi;
657
+ for (const m of commitMessage.matchAll(fileRe)) {
658
+ // Only count values that look like file paths (have an extension or a slash)
659
+ const val = m[1];
660
+ if (/[./]/.test(val) && !val.includes(' ')) {
661
+ claims.push({ kind: 'file', value: val, raw: m[0] });
662
+ }
663
+ }
664
+
665
+ // Dedup
666
+ const seen = new Set();
667
+ return {
668
+ claims: claims.filter(c => {
669
+ const k = `${c.kind}::${c.value.toLowerCase()}`;
670
+ if (seen.has(k)) return false;
671
+ seen.add(k);
672
+ return true;
673
+ })
674
+ };
675
+ }
676
+
677
+ /**
678
+ * Check commit message claims against the staged diff. Each claim must appear
679
+ * somewhere in the diff (a file path in the changed-files list OR the token
680
+ * appearing as-is in the diff body).
681
+ *
682
+ * @param {string} commitMessage
683
+ * @param {Object} [opts]
684
+ * @param {string} [opts.diffText] — raw `git diff --staged` output
685
+ * @param {string[]} [opts.changedFiles] — staged file list (alternative input)
686
+ * @returns {{ok: boolean, totalClaims: number, missingClaims: Array, verifiedClaims: Array}}
687
+ */
688
+ function verifyCommitMessageAgainstDiff(commitMessage, opts = {}) {
689
+ const { claims } = parseCommitMessageClaims(commitMessage);
690
+ if (claims.length === 0) return { ok: true, totalClaims: 0, missingClaims: [], verifiedClaims: [] };
691
+
692
+ const diffText = typeof opts.diffText === 'string' ? opts.diffText : '';
693
+ const changedFiles = Array.isArray(opts.changedFiles) ? opts.changedFiles : [];
694
+ const haystack = [diffText, ...changedFiles].join('\n');
695
+
696
+ const missingClaims = [];
697
+ const verifiedClaims = [];
698
+
699
+ for (const claim of claims) {
700
+ let found = false;
701
+ if (claim.kind === 'file') {
702
+ // File claims verify by exact path match (or suffix) in changed-files list
703
+ found = changedFiles.some(f => f === claim.value || f.endsWith('/' + claim.value) || f.endsWith(claim.value));
704
+ if (!found) found = diffText.includes(claim.value);
705
+ } else {
706
+ // finding-id + task-id: plain substring search in the haystack
707
+ found = haystack.includes(claim.value);
708
+ }
709
+ (found ? verifiedClaims : missingClaims).push(claim);
710
+ }
711
+
712
+ return {
713
+ ok: missingClaims.length === 0,
714
+ totalClaims: claims.length,
715
+ missingClaims,
716
+ verifiedClaims
717
+ };
718
+ }
719
+
720
+ /**
721
+ * Human-readable message when claims are missing from the diff.
722
+ *
723
+ * @param {Object} result — from verifyCommitMessageAgainstDiff
724
+ * @returns {string|null}
725
+ */
726
+ function formatMissingClaimsMessage(result) {
727
+ if (!result || result.ok || !Array.isArray(result.missingClaims) || result.missingClaims.length === 0) {
728
+ return null;
729
+ }
730
+ const lines = [
731
+ `Commit message claims ${result.missingClaims.length} item(s) that do not appear in the staged diff:`
732
+ ];
733
+ for (const c of result.missingClaims) {
734
+ lines.push(` • ${c.kind === 'finding-id' ? 'Finding' : c.kind === 'task-id' ? 'Task' : 'File'} "${c.value}" — not found`);
735
+ }
736
+ lines.push('');
737
+ lines.push('Options:');
738
+ lines.push(' 1. Add the missing fix to the commit now (git add + amend)');
739
+ lines.push(' 2. Remove the unverified claim from the commit message');
740
+ lines.push(' 3. Acknowledge + proceed (use --force-commit-claims if blocking from a gate)');
741
+ return lines.join('\n');
742
+ }
743
+
617
744
  // ============================================================
618
745
  // Exports
619
746
  // ============================================================
@@ -627,6 +754,9 @@ module.exports = {
627
754
  isTruthGateDisabled,
628
755
  getMinTierForDone,
629
756
  scanForClaimContradictions,
757
+ parseCommitMessageClaims,
758
+ verifyCommitMessageAgainstDiff,
759
+ formatMissingClaimsMessage,
630
760
  TIER_NAMES,
631
761
  DONE_WORDS,
632
762
  DISAGREEMENT_WORDS,
package/scripts/flow-config-defaults.js CHANGED
@@ -818,6 +818,31 @@ const CONFIG_DEFAULTS = {
818
818
  // --- Gate Confidence ---
819
819
  gateConfidence: { enabled: false },
820
820
 
821
+ // --- Intent-Grounded Reasoning (IGR) ---
822
+ // Master flag for the IGR pipeline: Intent Framing (Step 1.15), Architect
823
+ // Pass (Step 1.55), Logic Adversary (Step 1.57), Scope-Confidence Audit
824
+ // (Step 1.45), Completion Truth Gate (Step 3.9). Default-on so new projects
825
+ // inherit the full reasoning pipeline. See .claude/docs/intent-grounded-reasoning.md.
826
+ intentGroundedReasoning: {
827
+ enabled: true,
828
+ _comment: 'IGR pipeline: architect + logic adversary + truth gate. See .claude/docs/intent-grounded-reasoning.md'
829
+ },
830
+
831
+ // --- Research Reasoning Gate ---
832
+ // Tiered classification for conversation-mode questions. Tier 1 = factual,
833
+ // direct answer. Tier 2 = domain/recommendation, surface assumptions and
834
+ // wait for user confirmation. Tier 3 = architecture, tier 2 flow + spawn
835
+ // cross-model adversary. See wogi-start.md § Research Reasoning Gate.
836
+ researchReasoningGate: {
837
+ enabled: true,
838
+ tier2: { enabled: true },
839
+ tier3: {
840
+ enabled: true,
841
+ adversaryModel: 'sonnet',
842
+ _comment_adversaryModel: 'Model used for Tier-3 cross-model adversary. Reused by /wogi-peer-review, /wogi-debug-hypothesis, /wogi-learn, /wogi-decide — single canonical key.'
843
+ }
844
+ },
845
+
821
846
  // --- Long Input Gate ---
822
847
  longInputGate: {
823
848
  enabled: true,
package/scripts/flow-extraction-review.js CHANGED
@@ -104,8 +104,13 @@ function loadReviewSession() {
104
104
  return null;
105
105
  }
106
106
 
107
- // Check for prototype pollution keys
108
- if ('__proto__' in parsed || 'constructor' in parsed || 'prototype' in parsed) {
107
+ // Check for prototype pollution keys. Use Object.prototype.hasOwnProperty
108
+ // rather than `key in parsed` — the latter also returns true for inherited
109
+ // properties, and EVERY plain object inherits `constructor` from
110
+ // Object.prototype, which made this guard falsely trip on every valid
111
+ // session file (pre-existing bug, found via v2.25.1 wave2 test).
112
+ const hasOwn = Object.prototype.hasOwnProperty;
113
+ if (hasOwn.call(parsed, '__proto__') || hasOwn.call(parsed, 'constructor') || hasOwn.call(parsed, 'prototype')) {
109
114
  console.error('Review session file contains unsafe keys');
110
115
  return null;
111
116
  }
@@ -414,11 +419,21 @@ function exportAsItemManifest() {
414
419
  // Coordinate with Intent Bootstrap (see flow-story-gates.coordinateIntentBootstrap)
415
420
  // so /wogi-start doesn't re-prompt if the user already scheduled bootstrap via
416
421
  // /wogi-story during this session.
422
+ //
423
+ // v2.25.1: Semantics corrected (nit from Waves 1-3 review). The flag
424
+ // represents "is IGR bootstrap active/scheduled for this session?", NOT
425
+ // "did THIS call schedule it?". `result.active` is true when IGR is enabled
426
+ // and bootstrap has been scheduled — whether by this call or a prior one.
417
427
  let intentBootstrapScheduled = false;
418
428
  try {
419
429
  const gates = require('./flow-story-gates');
420
430
  const result = gates.coordinateIntentBootstrap();
421
- intentBootstrapScheduled = !!(result && result.scheduled);
431
+ if (result && result.active) {
432
+ // Scheduled in this call OR already-scheduled from a prior call = active
433
+ intentBootstrapScheduled = result.scheduled === true ||
434
+ result.reason === 'already-scheduled' ||
435
+ result.reason === 'artifacts-exist';
436
+ }
422
437
  } catch (_err) { /* non-critical */ }
423
438
 
424
439
  return {
package/scripts/flow-morning.js CHANGED
@@ -385,22 +385,31 @@ function collectBriefingData() {
385
385
  // v2.23.0 — Workspace dispatch surfacing (manager mode only).
386
386
  // If the user is working inside a workspace manager session, surface any
387
387
  // overdue or restart-gap-lost dispatches so the morning briefing catches
388
- // what the last manager turn would have caught. Fail-open.
388
+ // what the last manager turn would have caught. Fail-open; DEBUG-logged.
389
389
  try {
390
390
  if (process.env.WOGI_WORKSPACE_ROOT) {
391
391
  const { buildOverdueContext } = require('./hooks/core/overdue-dispatches');
392
392
  const ctx = buildOverdueContext();
393
393
  if (ctx) briefing.workspaceOverdue = ctx;
394
394
  }
395
- } catch (_err) { /* non-critical */ }
395
+ } catch (err) {
396
+ if (process.env.DEBUG) {
397
+ console.error(`[morning] Workspace overdue check failed (fail-open): ${err.message}`);
398
+ }
399
+ }
396
400
 
397
401
  // v2.23.0 — Completion-claim honesty scan.
398
402
  // Catches done-word-in-notes-while-status-partial and similar
399
403
  // contradictions across ready.json (uses the honesty-infra from 2026-04-16).
404
+ // Fail-open; DEBUG-logged.
400
405
  try {
401
406
  const { checkCompletionClaimHonesty } = require('./flow-health');
402
407
  briefing.honestyHits = checkCompletionClaimHonesty();
403
- } catch (_err) { /* non-critical */ }
408
+ } catch (err) {
409
+ if (process.env.DEBUG) {
410
+ console.error(`[morning] Honesty scan failed (fail-open): ${err.message}`);
411
+ }
412
+ }
404
413
 
405
414
  // Generate suggested prompt if enabled
406
415
  if (morningConfig.generatePrompt !== false) {
package/scripts/flow-session-end.js CHANGED
@@ -596,9 +596,18 @@ function writeWorkspaceSessionEndMessage() {
596
596
  const workspaceRoot = process.env.WOGI_WORKSPACE_ROOT;
597
597
  if (!workspaceRoot) return;
598
598
  const repo = process.env.WOGI_REPO_NAME;
599
- // Only manager-mode sessions emit this signal. Workers use their own
600
- // Stop-hook worker-stopped message (see lib/workspace-messages.js).
601
- if (repo && repo !== 'manager') return;
599
+ // Only emit this signal from EXPLICIT manager-mode sessions.
600
+ // v2.25.1 (M2 from Waves 1-3 review): tightened to require
601
+ // WOGI_REPO_NAME === 'manager' explicitly. Previously we let
602
+ // unset-repo sessions fall through, which could emit a spurious
603
+ // "manager session ended" broadcast from a mis-env'd worker shell.
604
+ // Workers use their own Stop-hook worker-stopped message.
605
+ if (repo !== 'manager') {
606
+ if (repo && process.env.DEBUG) {
607
+ console.error(`[session-end] Skipping workspace message — WOGI_REPO_NAME is '${repo}', not 'manager'`);
608
+ }
609
+ return;
610
+ }
602
611
 
603
612
  try {
604
613
  const messagesLib = path.resolve(__dirname, '..', 'lib', 'workspace-messages.js');