maestro-flow 0.5.45 → 0.5.46

package/.agy/skills/odyssey-review-test-fix/SKILL.md

@@ -14,254 +14,339 @@ allowed-tools:
   - view_file
   - write_to_file
 ---
-<base>@~/.maestro/workflows/odyssey-base.md</base>
-
-<purpose>
-archaeology → explore → multi-dimensional review → fix ALL findings → confirm → generalize → discover → persist. Zero-residual: every finding gets an action.
-</purpose>
-
-<boundary>
-**In scope:** Multi-dimensional deep review of target code → exhaustive fix ALL findings by severity → generalize patterns project-wide.
-**Out of scope:** Root cause debug → `/odyssey-debug` | Feature implementation → `/odyssey-planex` | UI visual optimization → `/odyssey-ui`
-
-**Exploration freedom:** Free within boundary — cross-dimension correlation, git history tracing, project-wide generalization scan. Fix ALL findings within fix_threshold (default: all).
-**Zero-residual:** Every finding MUST have a concrete action (fix / issue / decision). "Report and shelve" and "pre-existing skip" are forbidden.
-</boundary>
-
-<context>
-$ARGUMENTS
-
-**Target resolution:**
-
-| Input | Resolution |
-|-------|-----------|
-| File/dir path | Review those files |
-| `HEAD` / `staged` | `git diff HEAD` / `git diff --staged` |
-| Phase number | state.json → changed files |
-| PR number | `git diff main...HEAD` |
-
-**Flags:** `--dimensions <list>` subset of review dimensions | `--fix-threshold <level>` severity cutoff (default: all) | `--skip-fix` skip S_FIX+S_CONFIRM | `--skip-generalize` skip S_GENERALIZE+S_DISCOVER | `--auto` no delegate confirmation | `-y` auto-confirm | `-c` resume | `--heartbeat` /loop heartbeat
-
-**Session**: `.workflow/scratch/{YYYYMMDD}-review-odyssey-{slug}/`
-**Output**: `session.json` | `evidence.ndjson` | `explore.json` | `understanding.md`
-
-**session.json — review-specific fields:**
-```json
-{ "target": "", "dimensions": [], "review_result": {"remaining_actionable": 0},
-  "patterns": [], "confirmation": null, "generalization_stats": null }
-```
-
-**evidence.ndjson phases:** `archaeology|explore|review|fix|discovery|decision|self-iteration`
-
-**phase_goals[]:**
-
-| ID | Goal | done_when | phase | skip_when |
-|----|------|-----------|-------|-----------|
-| G1 | Review completed | all dimensions reviewed | S_REVIEW | — |
-| G2 | Explore context | explore.json populated | S_EXPLORE | — |
-| G3 | Zero remaining | `remaining_actionable == 0` | S_CONFIRM | skip_fix |
-| G4 | Pattern generalized | patterns[] >=1 | S_GENERALIZE | skip_generalize |
-| G5 | Discoveries triaged | all hits classified | S_DISCOVER | skip_generalize |
-| G6 | Learnings persisted | spec entries or no actionable | S_RECORD | — |
-
-**understanding.md — 8 sections:**
-1. Target & Scope ← S_INTAKE | 2. Archaeology ← S_ARCHAEOLOGY | 3. Exploration ← S_EXPLORE
-4. Review Results ← S_REVIEW | 5. Fix & Confirmation ← S_FIX+S_CONFIRM
-6. Generalization ← S_GENERALIZE | 7. Discoveries ← S_DISCOVER | 8. Learnings ← S_RECORD
-
-Specs: `maestro load --type spec --category review`
-
-**Knowledge Persistence categories (section 8):**
-
-| Category | Content | Follow-up |
-|----------|---------|-----------|
-| Cross-dimension recurring pattern | Pattern + affected dimensions + coding standard | `/spec-add review` |
-| Security finding | Vulnerability type + triggers + fix approach | `/spec-add debug` |
-| Architecture violation pattern | Violation + correct boundary + verification | `/spec-add arch` |
-| Reusable generalization pattern | Signature + risk + fix template + scope | `/spec-add coding` |
-</context>
-
-<invariants>
-All invariants defined in base.
-</invariants>
-
-<self_iteration>
-Applies to: **S_ARCHAEOLOGY, S_EXPLORE, S_REVIEW, S_FIX, S_GENERALIZE**. Logic in base.
-</self_iteration>
-
-<state_machine>
-
-<states>
-S_INTAKE → S_ARCHAEOLOGY → S_EXPLORE → S_REVIEW → S_FIX → S_CONFIRM → S_GENERALIZE → S_DISCOVER → S_RECORD → END
-</states>
-
-<transitions>
-S_INTAKE → S_INTAKE       : -c + session found → A_RESUME_SESSION
-S_INTAKE → S_ARCHAEOLOGY  : target resolved → A_INTAKE
-S_INTAKE → S_INTAKE       : no target → ask_question
-
-S_ARCHAEOLOGY → S_EXPLORE     : complete
-S_EXPLORE     → S_REVIEW      : complete
-
-S_REVIEW  → S_FIX          : !skip_fix AND findings
-S_REVIEW  → S_GENERALIZE   : skip_fix OR no findings, !skip_generalize
-S_REVIEW  → S_RECORD       : both skip
-
-S_FIX     → S_CONFIRM      : tier complete
-S_CONFIRM → S_GENERALIZE   : confirmed, !skip_generalize
-S_CONFIRM → S_RECORD       : confirmed, skip_generalize
-S_CONFIRM → S_FIX          : needs_rework
-
-S_GENERALIZE → S_DISCOVER  : similar code found
-S_GENERALIZE → S_RECORD    : no similar code
-
-S_DISCOVER → S_FIX         : fixable sibling → cross_phase_loops++
-S_DISCOVER → S_REVIEW      : new target, loops < max_loops → cross_phase_loops++
-S_DISCOVER → S_RECORD      : remaining_actionable == 0 OR loops >= max_loops
-</transitions>
-
-<actions>
-
-### A_INTAKE
-1. Parse target + flags → file list. Create SESSION_DIR, derive phase_goals[]
-2. `maestro search "<keywords>"` + Glob prior sessions + ARCHITECTURE.md + Grep keywords
-3. Write `session.json` + `understanding.md` section 1, emit Goal Prompt
-
-Commit: `"odyssey-review({slug}): INTAKE — parse target and load context"`
-
-### A_RESUME_SESSION
-Glob latest session → read `session.json` → jump to `current_state`.
-
-### A_ARCHAEOLOGY
-2 parallel Agents: Timeline (`git log --oneline -20 -- {files}`) + Blame (top 3 files `git blame -L {region}`). Evidence phase=archaeology.
-
-**Error handling:** If any archaeology agent fails (Timeline or Blame), log W003 and proceed with available results. If delegate fails, log W003 and proceed with local agent results only.
-
-`maestro delegate --role analyze --mode analysis` (`run_in_background: true`):
-- PURPOSE: Review recent modifications related to {target}
-- EXPECTED: JSON [{commit_sha, risk_level, analysis, could_cause_issue, explanation}]
-
-Update section 2. Commit: `"odyssey-review({slug}): ARCHAEOLOGY — git history analysis"`
-
-### A_EXPLORE
-Skip if no CLI tools (W006).
-
-`maestro delegate --role explore --mode analysis` (`run_in_background: true`):
-- PURPOSE: Call chains, recent changes, error gaps, similar patterns
-- EXPECTED: JSON {call_chains, recent_changes, error_gaps, similar_patterns}
-
-Write `explore.json` + evidence phase=explore. Update section 3. Mark G2. Commit: `"odyssey-review({slug}): EXPLORE — codebase exploration"`
-
-### A_REVIEW
-Spawn N parallel Agents, one per dimension:
-- **Correctness**: logic errors, boundary conditions, null/undefined, race conditions
-- **Security**: injection, XSS, CSRF, data exposure, auth bypass
-- **Performance**: hot paths, N+1, memory leaks, unnecessary recomputation
-- **Architecture**: layer violations, circular deps, interface contracts, SoC
-
-Each returns `[{title, severity, file, line, description, suggestion, cwe}]`. Merge → evidence phase=review. Write `review_result` + section 4 severity matrix. Mark G1.
-
-Commit: `"odyssey-review({slug}): REVIEW — multi-dimension review complete"`
-
-### A_FIX
-Exhaustive iterative fix — descend by severity until `remaining_actionable == 0`.
-
-```
-for tier in [critical, high, medium, low].filter(>= threshold):
-  for each unfixed candidate: read +/-20 lines → fix → evidence phase=fix
-  re-review modified area (new findings → append, continue; max 2 per tier)
-  tier done → auto-commit
-```
-
-Normal: ask_question per tier. `-y`: auto-fix all.
-Remaining > 0 → retry (max_fix_rounds = 5). Unchanged 2 rounds → classify each individually. After 5 rounds with remaining > 0 → escalate: Normal: ask_question (continue/accept/reclassify) | `-y`: classify remaining as `deferred`, proceed.
-Blanket "pre-existing" forbidden.
-
-Commit per tier: `"odyssey-review({slug}): FIX-{tier} — {N} items fixed"`
-
-### A_CONFIRM
-Run tests + `maestro delegate --role review --mode analysis` (`run_in_background: true`) for zero-residual review.
-- `remaining == 0 AND new == 0` → confirmed, mark G3
-- Otherwise → needs_rework → S_FIX
-
-Update `confirmation` + `remaining_actionable` + section 5.
-
-Commit: `"odyssey-review({slug}): CONFIRM — zero-residual verified"`
-
-### A_GENERALIZE, A_DISCOVER, A_RECORD
-Base shared_actions. Review overrides:
-- **A_GENERALIZE** pattern source: findings with severity >= medium
-- **A_RECORD** learnings per Knowledge Persistence table
-
-**Completion summary:**
-```
---- REVIEW-TEST-FIX ODYSSEY COMPLETE ---
-Target:     {target}          Dimensions: {dims}
-Findings:   {C}C {H}H {M}M {L}L    Fix: {fixed}, confirmed={yes|skip}
-Patterns:   {N} ({by_layer})        Scan hits: {total} ({cross} cross-layer)
-Issues:     {N} created
-Decisions:  {N} resolved, {M} pending, {K} deferred
-Learnings:  {N} persisted
-Self-iter:  {N} rounds across {M} stages
-Goals:      {done}/{total} ({skipped} skipped)
----
-```
-
-</actions>
-
-<appendix>
-
-### `-y` review-specific points
-
-| Decision Point | Normal | `-y` |
-|---------------|--------|------|
-| S_FIX tier candidates | ask_question | auto-fix, deferred |
-| S_FIX re-review new findings | ask_question | auto-append |
-| S_CONFIRM needs_rework | Display → S_FIX | auto proceed |
-
-### Goal Prompt convergence rules
-
-```
-Stop when remaining_actionable == 0, confirmation == confirmed,
-generalization exhausted, phase_goals_all_done=true.
-Fix iterates by severity tier; each tier re-reviews modified area, new findings appended.
-Every finding must have action (fix/issue/decision). Decision pending must ask_question.
-```
-
-</appendix>
-
-</state_machine>
-
-<error_codes>
-| Code | Severity | Condition | Recovery |
-|------|----------|-----------|----------|
-| E001 | error | No target specified | Provide target |
-| E002 | error | Target path not found | Check path |
-| W001 | warning | No git history | Proceed |
-| W002 | warning | Some dimension agents failed | Partial coverage |
-| W003 | warning | Archaeology agent or delegate failure | Proceed with available results |
-</error_codes>
-
-<success_criteria>
-- [ ] Session + 4 output files + prior knowledge searched
-- [ ] Archaeology + CLI review → evidence phase=archaeology
-- [ ] CLI exploration → explore.json + evidence phase=explore
-- [ ] All dimensions reviewed, ALL findings fixed (remaining_actionable == 0)
-- [ ] Per-tier re-review gate; every unfixed finding individually classified
-- [ ] understanding.md sections 1-8 progressive
-- [ ] Fix + confirmed (unless --skip-fix)
-- [ ] Generalization + scan (unless --skip-generalize)
-- [ ] Discoveries classified; unfixed findings individually justified
-- [ ] phase_goals G1-G6 audited, Goal Prompt once, `-y` no blocking, -c resumable
-- [ ] Completion summary
-</success_criteria>
-
-<next_step_routing>
-| Condition | Next |
-|-----------|------|
-| Deeper debug needed | `/odyssey-debug "<finding>"` |
-| Issues created | `/manage-issue list --source review-odyssey` |
-| Document pattern | `/learn-decompose <module>` |
-| Plan fixes | `/maestro-plan --gaps` |
-</next_step_routing>
-</output>
+<base>@~/.maestro/workflows/odyssey-base.md</base>
+
+<purpose>
+archaeology → explore → multi-dimensional review → fix ALL findings → confirm → generalize → discover → persist. Zero-residual: every finding gets an action.
+</purpose>
+
+<boundary>
+**In scope:** Multi-dimensional deep review of target code → exhaustive fix ALL findings by severity → generalize patterns project-wide.
+**Out of scope:** Root cause debug → `/odyssey-debug` | Feature implementation → `/odyssey-planex` | UI visual optimization → `/odyssey-ui`
+
+**Exploration freedom:** Free within boundary — cross-dimension correlation, git history tracing, project-wide generalization scan. Fix ALL findings within fix_threshold (default: all).
+**Zero-residual:** Every finding MUST have a concrete action (fix / issue / decision). "Report and shelve" and "pre-existing skip" are forbidden.
+</boundary>
+
+<context>
+$ARGUMENTS
+
+**Target resolution:**
+
+| Input | Resolution |
+|-------|-----------|
+| File/dir path | Review those files |
+| `HEAD` / `staged` | `git diff HEAD` / `git diff --staged` |
+| Phase number | state.json → changed files |
+| PR number | `git diff main...HEAD` |
+
+**Flags:** `--dimensions <list>` subset of review dimensions | `--fix-threshold <level>` severity cutoff (default: all) | `--skip-fix` skip S_FIX+S_CONFIRM | `--skip-generalize` skip S_GENERALIZE+S_DISCOVER | `--auto` no delegate confirmation | `-y` auto-confirm | `-c` resume | `--heartbeat` /loop heartbeat
+
+**Session**: `.workflow/scratch/{YYYYMMDD}-review-odyssey-{slug}/`
+**Output**: `session.json` | `evidence.ndjson` | `explore.json` | `understanding.md`
+
+**session.json — review-specific fields:**
+```json
+{ "target": "", "dimensions": [], "review_result": {"remaining_actionable": 0},
+  "patterns": [], "confirmation": null, "generalization_stats": null }
+```
+
+**evidence.ndjson phases:** `archaeology|explore|review|fix|discovery|decision|self-iteration`
+
+**phase_goals[]:**
+
+| ID | Goal | done_when | phase | skip_when |
+|----|------|-----------|-------|-----------|
+| G1 | Review completed | all dimensions reviewed | S_REVIEW | — |
+| G2 | Explore context | explore.json populated | S_EXPLORE | — |
+| G3 | Zero remaining | `remaining_actionable == 0` | S_CONFIRM | skip_fix |
+| G4 | Pattern generalized | patterns[] >=1 | S_GENERALIZE | skip_generalize |
+| G5 | Discoveries triaged | all hits classified | S_DISCOVER | skip_generalize |
+| G6 | Learnings persisted | spec entries or no actionable | S_RECORD | — |
+
+**understanding.md — 8 sections:**
+1. Target & Scope ← S_INTAKE | 2. Archaeology ← S_ARCHAEOLOGY | 3. Exploration ← S_EXPLORE
+4. Review Results ← S_REVIEW | 5. Fix & Confirmation ← S_FIX+S_CONFIRM
+6. Generalization ← S_GENERALIZE | 7. Discoveries ← S_DISCOVER | 8. Learnings ← S_RECORD
+
+Specs: `maestro load --type spec --category review`
+
+**Knowledge Persistence categories (section 8):**
+
+| Category | Content | Follow-up |
+|----------|---------|-----------|
+| Cross-dimension recurring pattern | Pattern + affected dimensions + coding standard | `/spec-add review` |
+| Security finding | Vulnerability type + triggers + fix approach | `/spec-add debug` |
+| Architecture violation pattern | Violation + correct boundary + verification | `/spec-add arch` |
+| Reusable generalization pattern | Signature + risk + fix template + scope | `/spec-add coding` |
+</context>
+
+<invariants>
+All invariants defined in base.
+6. **Generalize is mandatory** — S_GENERALIZE and S_DISCOVER execute unless `skip_generalize == true`. "Zero remaining" or context pressure are NOT valid reasons to skip. The phase itself determines whether patterns exist.
+</invariants>
+
+<self_iteration>
+Applies to: **S_ARCHAEOLOGY, S_EXPLORE, S_REVIEW, S_FIX, S_GENERALIZE**. Logic in base.
+</self_iteration>
+
+<state_machine>
+
+<states>
+S_INTAKE → S_ARCHAEOLOGY → S_EXPLORE → S_REVIEW → S_FIX → S_CONFIRM → S_GENERALIZE → S_DISCOVER → S_RECORD → END
+</states>
+
+<transitions>
+S_INTAKE → S_INTAKE       : -c + session found → A_RESUME_SESSION
+S_INTAKE → S_ARCHAEOLOGY  : target resolved → A_INTAKE
+S_INTAKE → S_INTAKE       : no target → ask_question
+
+S_ARCHAEOLOGY → S_EXPLORE     : complete
+S_EXPLORE     → S_REVIEW      : complete
+
+S_REVIEW  → S_FIX          : !skip_fix AND findings
+S_REVIEW  → S_GENERALIZE   : skip_fix OR no findings, !skip_generalize
+S_REVIEW  → S_RECORD       : both skip
+
+S_FIX     → S_CONFIRM      : tier complete
+S_CONFIRM → S_GENERALIZE   : confirmed, !skip_generalize
+S_CONFIRM → S_RECORD       : confirmed, skip_generalize
+S_CONFIRM → S_FIX          : needs_rework
+
+S_GENERALIZE → S_DISCOVER  : similar code found
+S_GENERALIZE → S_RECORD    : all 3 layers scanned with evidence, total_hits == 0
+
+S_DISCOVER → S_FIX         : fixable sibling → cross_phase_loops++
+S_DISCOVER → S_REVIEW      : new target, loops < max_loops → cross_phase_loops++
+S_DISCOVER → S_RECORD      : remaining_actionable == 0 OR loops >= max_loops
+</transitions>
+
+<actions>
+
+### A_INTAKE
+1. Parse target + flags → file list. Create SESSION_DIR, derive phase_goals[]
+2. `maestro search "<keywords>"` + Glob prior sessions + ARCHITECTURE.md + Grep keywords
+3. Write `session.json` + `understanding.md` section 1, emit Goal Prompt
+
+Commit: `"odyssey-review({slug}): INTAKE — parse target and load context"`
+
+### A_RESUME_SESSION
+Glob latest session → read `session.json` → jump to `current_state`.
+
+### A_ARCHAEOLOGY
+2 parallel Agents: Timeline (`git log --oneline -20 -- {files}`) + Blame (top 3 files `git blame -L {region}`). Evidence phase=archaeology.
+
+**Error handling:** If any archaeology agent fails (Timeline or Blame), log W003 and proceed with available results. If delegate fails, log W003 and proceed with local agent results only.
+
+`maestro delegate --role analyze --mode analysis` (`run_in_background: true`):
+- PURPOSE: Review recent modifications related to {target}
+- EXPECTED: JSON [{commit_sha, risk_level, analysis, could_cause_issue, explanation}]
+
+Update section 2. Commit: `"odyssey-review({slug}): ARCHAEOLOGY — git history analysis"`
+
+### A_EXPLORE
+Skip if no CLI tools (W006).
+
+`maestro delegate --role explore --mode analysis` (`run_in_background: true`):
+- PURPOSE: Call chains, recent changes, error gaps, similar patterns
+- EXPECTED: JSON {call_chains, recent_changes, error_gaps, similar_patterns}
+
+Write `explore.json` + evidence phase=explore. Update section 3. Mark G2. Commit: `"odyssey-review({slug}): EXPLORE — codebase exploration"`
+
+### A_REVIEW
+Spawn N parallel Agents, one per dimension:
+- **Correctness**: logic errors, boundary conditions, null/undefined, race conditions
+- **Security**: injection, XSS, CSRF, data exposure, auth bypass
+- **Performance**: hot paths, N+1, memory leaks, unnecessary recomputation
+- **Architecture**: layer violations, circular deps, interface contracts, SoC
+
+Each returns `[{title, severity, file, line, description, suggestion, cwe}]`. Merge → evidence phase=review. Write `review_result` + section 4 severity matrix. Mark G1.
+
+Commit: `"odyssey-review({slug}): REVIEW — multi-dimension review complete"`
+
+### A_FIX
+Exhaustive iterative fix — descend by severity until `remaining_actionable == 0`.
+
+```
+for tier in [critical, high, medium, low].filter(>= threshold):
+  for each unfixed candidate: read +/-20 lines → fix → evidence phase=fix
+  re-review modified area (new findings → append, continue; max 2 per tier)
+  tier done → auto-commit
+```
+
+Normal: ask_question per tier. `-y`: auto-fix all.
+Remaining > 0 → retry (max_fix_rounds = 5). Unchanged 2 rounds → classify each individually. After 5 rounds with remaining > 0 → escalate: Normal: ask_question (continue/accept/reclassify) | `-y`: classify remaining as `deferred`, proceed.
+Blanket "pre-existing" forbidden.
+
+Commit per tier: `"odyssey-review({slug}): FIX-{tier} — {N} items fixed"`
+
+### A_CONFIRM
+Run tests + `maestro delegate --role review --mode analysis` (`run_in_background: true`) for zero-residual review.
+- `remaining == 0 AND new == 0` → confirmed, mark G3
+- Otherwise → needs_rework → S_FIX
+
+Update `confirmation` + `remaining_actionable` + section 5.
+
+Commit: `"odyssey-review({slug}): CONFIRM — zero-residual verified"`
+
+### A_GENERALIZE
+
+**MANDATORY — executes unless `skip_generalize == true`. Prior-phase convergence, "zero remaining," or context pressure are NOT valid skip reasons.**
+
+Pattern source: review findings with severity >= medium.
+
+**Step 1 — 3-layer pattern extraction** from confirmed findings:
+
+| Layer | Method | Targets |
+|-------|--------|---------|
+| Syntax | Build regex from fix diffs → Grep | Unchecked return values, deprecated API calls, missing type assertion, unhandled await |
+| Semantic | Understand anti-pattern per dimension → Agent scan | Logic inversion in edge case, unhandled rejection, incomplete state machine transition |
+| Structural | Find files with same handler/validator shape | Similar route handlers, parallel DTO validators, analogous test fixtures |
+
+Write `session.json.patterns[]`: `[{id, source, layer, signature, description, risk, fix_template, confidence}]`
+
+**Thoroughness floor:** ALL 3 layers must be attempted and logged. Each layer records search method, scope, hit count in evidence phase=generalization. "No hits" requires all 3 layers to return 0 with logged evidence.
+
+**Step 2 — 4-agent concurrent scan** (single message, 4 Agents):
+
+| Agent | Strategy | Scope |
+|-------|----------|-------|
+| Syntax grep | Grep regex from pattern signatures | Full project |
+| Semantic scan | Per-dimension anti-pattern check | Related modules |
+| Structural match | Find structurally similar files | Full project |
+| Historical grep | `git log -S` for pattern signatures | Git history |
+
+**Step 3 — Cross-layer dedup:** multi-layer hit → boost | single-layer → `needs_review` | historically fixed → `regression_risk`
+
+**Step 4 — Iterative deepening:** Module with ≥3 hits → targeted deep scan (max 1 round).
+
+**Step 5 — Persist:** Update understanding.md section 6 + write `session.json.generalization_stats`:
+```json
+{"patterns_extracted": 0, "total_hits": 0, "cross_layer_confirmed": 0, "regression_risks": 0, "by_layer": {"syntax": 0, "semantic": 0, "structural": 0}, "deepening_triggered": false}
+```
+
+**Transition guard:** `S_GENERALIZE → S_RECORD` requires `by_layer` has entries for all 3 layers with search evidence logged. Mark G4 done.
+
+Commit: `"odyssey-review({slug}): GENERALIZE — pattern scan complete"`
+
+### A_DISCOVER
+
+**Executes whenever `total_hits > 0`. Cannot be skipped without `skip_generalize == true`.**
+
+1. **Triage** each hit with ±10 lines context → classify:
+   - `bug` — same defect pattern confirmed in sibling code
+   - `risk` — potential issue needing guard
+   - `safe` — false positive (must log individual reason — blanket "pre-existing" forbidden)
+
+2. **Route:**
+
+   | Classification | Action |
+   |---------------|--------|
+   | fixable sibling | Immediate fix → back to S_FIX, `cross_phase_loops++` |
+   | new target needing review | Route to S_REVIEW, `cross_phase_loops++` |
+   | risk + guard addable | Fix directly |
+   | risk + complex | Create issue |
+   | safe | Skip with logged per-item reason |
+
+   Normal: ask_question per hit | `-y`: auto-fix with template, create issue for rest
+
+3. **Cross-phase loops:** `loops >= max_loops` → must log per-item reasons, advance to S_RECORD.
+
+4. Append evidence phase=discovery. Update understanding.md section 7. Mark G5 done.
+
+Commit: `"odyssey-review({slug}): DISCOVER — sibling triage complete"`
+
+### A_RECORD
+
+1. Finalize understanding.md section 8 — learnings by Knowledge Persistence categories:
+   - Cross-dimension recurring pattern → `/spec-add review`
+   - Security finding → `/spec-add debug`
+   - Architecture violation pattern → `/spec-add arch`
+   - Reusable generalization pattern → `/spec-add coding`
+
+2. Mark G6 done. Pending decisions: Normal → ask_question | `-y` → skip (show deferred count).
+
+3. **Goal audit (hardened):**
+   - `done` → confirmed
+   - `skipped` → confirmed ONLY if corresponding `skip_when` flag is true
+   - **Hard rule:** G4 and G5 CANNOT be `skipped` unless `skip_generalize == true`. Pending without flag → `failed` (Normal: ask_question | `-y`: record `failed`)
+   - `phase_goals_all_done = true` only when all goals pass this audit
+
+4. `current_state = "COMPLETED"`, emit completion summary.
+
+**Completion summary:**
+```
+--- REVIEW-TEST-FIX ODYSSEY COMPLETE ---
+Target:     {target}          Dimensions: {dims}
+Findings:   {C}C {H}H {M}M {L}L    Fix: {fixed}, confirmed={yes|skip}
+Patterns:   {N} ({by_layer})        Scan hits: {total} ({cross} cross-layer)
+Issues:     {N} created
+Decisions:  {N} resolved, {M} pending, {K} deferred
+Learnings:  {N} persisted
+Self-iter:  {N} rounds across {M} stages
+Goals:      {done}/{total} ({skipped} skipped)
+---
+```
+
+Commit: `"odyssey-review({slug}): RECORD — summary and knowledge persistence"`
+
+</actions>
+
+<appendix>
+
+### `-y` review-specific points
+
+| Decision Point | Normal | `-y` |
+|---------------|--------|------|
+| S_FIX tier candidates | ask_question | auto-fix, deferred |
+| S_FIX re-review new findings | ask_question | auto-append |
+| S_CONFIRM needs_rework | Display → S_FIX | auto proceed |
+| A_DISCOVER hit routing | ask_question | auto-fix with template, create issue for rest |
+
+### Goal Prompt convergence rules
+
+```
+Stop when remaining_actionable == 0, confirmation == confirmed,
+generalization exhausted, phase_goals_all_done=true.
+Fix iterates by severity tier; each tier re-reviews modified area, new findings appended.
+Every finding must have action (fix/issue/decision). Decision pending must ask_question.
+```
+
+</appendix>
+
+</state_machine>
+
+<error_codes>
+| Code | Severity | Condition | Recovery |
+|------|----------|-----------|----------|
+| E001 | error | No target specified | Provide target |
+| E002 | error | Target path not found | Check path |
+| W001 | warning | No git history | Proceed |
+| W002 | warning | Some dimension agents failed | Partial coverage |
+| W003 | warning | Archaeology agent or delegate failure | Proceed with available results |
+| W004 | warning | Generalization 0 hits after full 3-layer scan | Advance to S_RECORD (requires all 3 layers attempted with evidence) |
+</error_codes>
+
+<success_criteria>
+- [ ] Session + 4 output files + prior knowledge searched
+- [ ] Archaeology + CLI review → evidence phase=archaeology
+- [ ] CLI exploration → explore.json + evidence phase=explore
+- [ ] All dimensions reviewed, ALL findings fixed (remaining_actionable == 0)
+- [ ] Per-tier re-review gate; every unfixed finding individually classified
+- [ ] understanding.md sections 1-8 progressive
+- [ ] Fix + confirmed (unless --skip-fix)
+- [ ] Generalization + scan (unless --skip-generalize)
+- [ ] Discoveries classified; unfixed findings individually justified
+- [ ] phase_goals G1-G6 audited, Goal Prompt once, `-y` no blocking, -c resumable
+- [ ] Completion summary
+</success_criteria>
+
+<next_step_routing>
+| Condition | Next |
+|-----------|------|
+| Deeper debug needed | `/odyssey-debug "<finding>"` |
+| Issues created | `/manage-issue list --source review-odyssey` |
+| Document pattern | `/learn-decompose <module>` |
+| Plan fixes | `/maestro-plan --gaps` |
+</next_step_routing>
+</output>