[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M6MZ1M776WHM38RCG355DTP6":3,"skill-same:01M6MZ1M776WHM38RCG355DTP6":72},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":31,"linkOnly":32,"markdownContent":33,"markdownBody":34,"frontMatter":35,"refName":37,"commitSha":38,"skillDirectory":39,"forksCount":40,"repositoryHtmlUrl":41,"repositoryHomepageUrl":6,"ownerHtmlUrl":42,"ownerBlogUrl":6,"ownerDescription":43,"updatedAt":44,"install":45,"resources":57},"01M6MZ1M776WHM38RCG355DTP6","execute-aa-ma-full",null,"Execute complete AA-MA plan from current position to completion with automated checkpoints. Run only when the user asks to run a whole AA-MA plan.","snewhouse\u002Faa-ma-forge","snewhouse","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F6317143?v=4","01M6MYYWF28C4KBQHHWB9SC6TS","Stephen J Newhouse","01M6MYYWJWQMDN2ZPFMJ584SSM",6,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fsnewhouse\u002Faa-ma-forge\u002Fblob\u002Fmain\u002Fclaude-code\u002Fskills\u002Fexecute-aa-ma-full\u002FSKILL.md",[],[19,20,21,22,23,24,25,26,27,28,29,30],"execute","full","execute aa ma full","run","plan","user","only","complete","position","with","checkpoints","true","2026-11-10 10:12:30",false,"---\nname: execute-aa-ma-full\ndescription: Execute complete AA-MA plan from current position to completion with automated checkpoints. Run only when the user asks to run a whole AA-MA plan.\ndisable-model-invocation: true\n---\n\n# AA-MA Full Plan Execution\n\nYou are executing the **complete AA-MA plan** from current position to final completion.\n\n**Scope**: All remaining milestones in tasks.md\n**Validation**: Strict at every milestone boundary\n**Git**: Auto-commit + provenance at each milestone, final tag on completion\n\n⚠️ **Use this command only for well-understood plans that don't need frequent human checkpoints.**\n\n---\n\n## 1. Locate Active Task & Pre-Execution Validation\n\nSee [pre-execution-validation.md](references\u002Fpre-execution-validation.md#1-locate-active-task--pre-execution-validation) — the Tier 1 existence check and the Tier 2 validator dispatch, with auto-recovery.\n\n---\n\n## 2. Full Context Injection\n\nSee [context-and-goal.md](references\u002Fcontext-and-goal.md#2-full-context-injection) — which AA-MA files to load for a full run, and the optional \u002Fgoal synthesis and bind for AFK runs.\n\n---\n\n## 2.5 Goal Synthesis & Bind (Optional, Recommended for AFK Runs)\n\nSee [context-and-goal.md](references\u002Fcontext-and-goal.md#25-goal-synthesis--bind-optional-recommended-for-afk-runs) — which AA-MA files to load for a full run, and the optional \u002Fgoal synthesis and bind for AFK runs.\n\n---\n\n## 3. Validate Loaded Context\n\n**Required checks**:\n- ✅ All 5 AA-MA files exist and are readable\n- ✅ REFERENCE contains key constants\n- ✅ TASKS has clear HTP structure with Status fields\n- ✅ At least one milestone has `Status: PENDING` or `Status: ACTIVE`\n- ✅ All milestones have explicit Acceptance Criteria defined\n- ✅ Dependencies between milestones are clearly specified\n\n**If validation fails**:\n```\nERROR: [Specific issue]\n- Missing files → \"AA-MA files incomplete. Run \u002Faa-ma-plan or check directory.\"\n- Malformed HTP → \"tasks.md lacks proper HTP format.\"\n- No pending milestones → \"All milestones complete.\"\n- Missing acceptance criteria → \"One or more milestones lack acceptance criteria.\"\n- Unclear dependencies → \"Milestone dependencies not specified.\"\n```\n\n---\n\n## 4. Parse HTP & Create TodoWrite Todos\n\n**Parse tasks.md for complete HTP structure**:\n- `## headers` = Milestones\n- `### headers` = Sub-tasks\n- Extract all metadata: `Status:`, `Dependencies:`, `Complexity:`, `Acceptance Criteria:`\n\n**Identify execution plan**:\n1. Count total milestones\n2. Count PENDING\u002FACTIVE milestones\n3. Count COMPLETE milestones\n4. Identify first PENDING milestone (starting point)\n5. Map dependency tree\n\n**Create TodoWrite todos for ALL remaining work**:\n- All PENDING\u002FACTIVE milestones\n- All sub-tasks within those milestones\n\n**Todo format**:\n```json\n{\n  \"content\": \"[Milestone\u002FTask title]\",\n  \"status\": \"pending|in_progress|completed\",\n  \"activeForm\": \"[Present continuous form]\"\n}\n```\n\n**Hierarchical organization**:\n```\n- [ ] Milestone 1: [title]\n  - [ ] Sub-task 1.1\n  - [ ] Sub-task 1.2\n- [ ] Milestone 2: [title]\n  - [ ] Sub-task 2.1\n  - [ ] Sub-task 2.2\n```\n\n---\n\n## 4.5 Quality Standards (Actionable Triggers)\n\nThese principles apply DURING execution. Each has a trigger condition and skip condition.\n\n- **KISS:** If implementation exceeds 200 lines for a single function or method, decompose before proceeding. Skip for generated code or data tables.\n- **DRY:** Before writing a new utility function, grep the codebase for existing implementations. If a match exists, reuse it. Section B.5 (Post-Milestone Simplification Review) catches misses post-hoc.\n- **SOLID:** If a class has >5 public methods, evaluate single-responsibility. If a module mixes data access + business logic + presentation, separate. Skip for scripts and one-off tools.\n- **SOC:** If a file grows beyond one clear responsibility, split. If a function handles both data transformation and side effects, separate.\n- **TDD:** If this milestone produces code, invoke `superpowers:test-driven-development`. Let the skill decide test strategy. Skip it only with a canonical `TDD-Waiver:` value (`engineering-standards.md` §2).\n- **12-Factor:** If the task involves service deployment, API servers, or containerized applications, reference 12-Factor principles for config (env vars not files), statelessness, and port binding. For ALL tasks with `.env` files, verify env-var-drift compliance (see `rules\u002Fenv-var-drift.md`).\n- **Context7 MCP:** Use for library docs and code generation. Retry once on failure, then fall back to WebSearch + official docs.\n- **WebSearch:** Fallback when Context7 fails. Also use proactively for unfamiliar APIs, recent library changes, or deployment patterns.\n\n### Pre-Execution Check (First Milestone Only)\n\nIf this is the **first milestone** and it touches 3+ files or unfamiliar code, invoke `Skill(system-mapping)` for the 5-point pre-flight check. Skip for subsequent milestones.\n\n---\n\n## 5. Execute Plan Sequentially\n\n**For each PENDING milestone (in order)**:\n\n### 5.1 Pre-Flight Checks\n\n1. **Update status**:\n   - Set milestone `Status: PENDING` → `Status: ACTIVE`\n   - Update TodoWrite milestone todo to `in_progress`\n\n2. **Dependency advisory — never blocks** (map Ticket 16): run the fence from\n   `\u002Fexecute-aa-ma-milestone` §5.1 step 3 (`aa_ma_deps advisory \"${TASKS_MD}\"`), show the\n   user any line it prints, log it to provenance.log, and continue.\n\n### 5.2 Execute All Sub-Tasks\n\n**For each sub-task in milestone**:\n1. Mark TodoWrite sub-task as `in_progress`\n1.5. **Mode Dispatch**: Parse `Mode:` from sub-task (inherit from milestone if absent, default HITL). HITL: display summary + criteria, `AskUserQuestion` [Proceed\u002FSkip\u002FAbort]. AFK: execute directly without pause.\n2. Execute sub-task (use agents, skills, Context7 MCP as needed; before editing existing code, run `Skill(impact-analysis)` on it)\n3. Verify result (guidance-based, trust your assessment)\n4. Update `Result Log:` in tasks.md\n5. Mark TodoWrite sub-task as `completed`\n\n### 5.3 Milestone Boundary Validation (STRICT)\n\n**REQUIRED validation before marking milestone COMPLETE:**\n\n#### A. Acceptance Criteria Verification\n- Explicitly verify EACH criterion met\n- Document verification method\u002Fresult for each\n- If ANY fails → set `Status: BLOCKED`, HALT, alert user\n\n#### B. Dependency Verification (for next milestone)\n- Advisory only: the next milestone's `Dependencies:` are reported by its own §5.1 step 2\n  when it becomes ACTIVE. Never halt on them.\n\n#### C. Impact Analysis Verification (REQUIRED)\n\n**Before completing milestone, verify no breaking changes introduced:**\n\n1. **Invoke the impact-analysis skill** (`Skill(impact-analysis)`)\n2. **Output consolidated impact analysis** for ALL files modified in this milestone\n3. **Verify no unresolved cascade effects**\n\n**Required output format** (consolidated for milestone):\n```\n? Impact Analysis: Milestone \"[Milestone Title]\"\n┌─ Files Modified: [N]\n│\n├─ [file_path]\n│  ├─ Upstream: [N] callers\n│  ├─ Contract: [YES\u002FNO]\n│  └─ Risk: [LOW\u002FMEDIUM\u002FHIGH]\n│\n└─ Overall Risk: [LOW\u002FMEDIUM\u002FHIGH]\n```\n\n**Validation rules**:\n- Overall Risk = LOW → Proceed to Test Execution\n- Overall Risk = MEDIUM → Document cascade updates, then proceed\n- Overall Risk = HIGH → HALT, present options (auto-fix, manual review, abort)\n\n**Never complete milestone with unresolved HIGH risk impacts.**\n\n#### D. Test Execution (if specified)\n- Run ALL listed tests\n- ALL must pass (zero failures)\n- Document results in Result Log\n- If failures → set `Status: BLOCKED`, HALT, alert user\n\n### 5.4 Finalization Protocol — MANDATORY\n\nBefore marking the milestone COMPLETE and creating git checkpoint, execute this 4-step finalization protocol at EACH milestone boundary. **No exceptions.**\n\n#### A. Integrity Check (Checklist Verification)\n\nDisplay acceptance criteria verification to the user:\n\n```\nAcceptance Criteria Verification:\n- ✓ [Criterion 1]: Confirmed - [brief evidence]\n- ✓ [Criterion 2]: Confirmed - [brief evidence]\n\nGate: [HARD|SOFT] — [Approval artifact found \u002F Convention-based]\nAll [X] criteria verified. Ready for finalization.\n```\n\n**Rules:**\n- Every acceptance criterion must be explicitly listed with evidence\n- If ANY criterion cannot be confirmed → HALT, do not proceed\n- If `Gate: HARD` and no approval artifact in context-log.md → HALT, request approval (see `\u002Fexecute-aa-ma-milestone` Section 7.1)\n- If `[task]-tests.yaml` exists, auto-run tests for this milestone (see `\u002Fexecute-aa-ma-milestone` Section 6.4)\n\n#### B. Documentation Auto-Update\n\nAutomatically update all 5 AA-MA files:\n\n| File | Auto-Update Action |\n|------|-------------------|\n| `tasks.md` | Mark milestone `Status: COMPLETE`, fill `Result Log:` |\n| `reference.md` | Add any new immutable facts discovered |\n| `context-log.md` | Append completion summary |\n| `provenance.log` | Append completion entry (after commit) |\n| `plan.md` | No change (historical record) |\n\n#### B.5. Post-Milestone Simplification Review\n\n**Skip if:** Only docs\u002Fconfig changed, diff \u003C 20 lines code, or `--skip-simplify` flag passed to `\u002Fexecute-aa-ma-full`.\n\nLaunch 3 parallel agents against milestone diff (reuse, quality, efficiency). CRITICAL findings require acknowledgment at approval gate. See `\u002Fexecute-aa-ma-milestone` Section 6.6 for full protocol.\n\n**Note:** For fully automated AFK runs, pass `--skip-simplify` to skip this review at every milestone and save tokens. The review is most valuable when running milestones interactively.\n\n**On failure:** Skip review, log to provenance. Do NOT block finalization.\n\n#### B.6. Post-Impl Adversarial Review (Phase 6.8 — NEW in v0.8.0)\n\nReference: ADR-0005 + `\u002Fexecute-aa-ma-milestone` Section 6.8 (full anatomy).\n\n**Skip if:** plan's `Created:` date is \u003C v0.8.0 release tag commit date (grandfathered), OR `AA_MA_AUDIT_BUDGET=off`, OR `AA_MA_HOOKS_DISABLE=1`.\n\nDelegate to `\u002Fexecute-aa-ma-milestone` Section 6.8 — same logic, same agents:\nthe `verify-impl` skill orchestrator dispatches up to 5 audit agents\n(code-reviewer, security-auditor, tdd-sequence-auditor,\ncontext7-evidence-auditor, future-proofing-auditor) per the milestone's\nplan-declared `Audit-Profile:` field. Aggregated verdict written to\n`[task]-impl-review.md`.\n\n**Verdicts:**\n- `PASS` \u002F `PASS_WITH_WARNINGS` → proceed to C (User Authorization)\n- `BLOCKED` (any CRITICAL accepted via override panel) → HALT this milestone, surface remediation to user. Do NOT auto-advance to next milestone. The user must fix the accepted CRITICAL, re-run `\u002Fexecute-aa-ma-milestone` (which re-runs §6.8), and only when §6.8 passes clean can `\u002Fexecute-aa-ma-full` resume.\n\n**Provenance:** §6.8 entry emitted by the milestone command itself. No duplicate entry from the full command — the delegation is logical, not physical.\n\n#### C. User Authorization (Approval Gate)\n\nUse AskUserQuestion to get explicit user approval:\n\n```\n? Finalization Review\n\nMilestone: [Title]\nAcceptance Criteria: [X\u002FX] verified ✓\n\nReady to mark COMPLETE?\n- Approve: Proceed with commit\n- Review First: Show detailed verification\n- Reject: Keep as ACTIVE\n```\n\n**Behavior:**\n- **Approve**: Proceed to git commit\n- **Review First**: Display full checklist, then re-prompt\n- **Reject**: HALT execution, keep `Status: ACTIVE`, ask user what needs fixing\n\n#### D. Transparent Status Change\n\nAfter approval, display minimal confirmation:\n\n```\n✅ Milestone marked COMPLETE: [Title]\n```\n\nThen proceed to git commit.\n\n### 5.5 Git Checkpoint Creation\n\n**If validation passes:**\n\n```bash\n# Stage changes\ngit add .\n\n# Get milestone info\nTASK_NAME=\"[task-name]\"\nMILESTONE_TITLE=\"[milestone title]\"\nMILESTONE_ID=\"[milestone-id]\"\nCOMPLEXITY=\"[complexity %]\"\n\n# Create structured commit\ngit commit -m \"feat($TASK_NAME): Complete $MILESTONE_TITLE\n\nMilestone: $MILESTONE_ID\nComplexity: $COMPLEXITY%\n\nAcceptance criteria verified:\n- [criterion 1]: ✓ [method]\n- [criterion 2]: ✓ [method]\n\nTasks completed:\n- [sub-task 1]\n- [sub-task 2]\n\nTests: [status]\n\n[AA-MA Plan] $TASK_NAME .claude\u002Fdev\u002Factive\u002F$TASK_NAME\"\n\n# Update provenance log\nCOMMIT_HASH=$(git rev-parse --short HEAD)\nTIMESTAMP=$(date -Iseconds)\necho \"[$TIMESTAMP] Commit $COMMIT_HASH — MilestoneID: $MILESTONE_ID — Status: COMPLETE\" >> .claude\u002Fdev\u002Factive\u002F$TASK_NAME\u002F${TASK_NAME}-provenance.log\n\n# Commit provenance update\ngit add .claude\u002Fdev\u002Factive\u002F$TASK_NAME\u002F${TASK_NAME}-provenance.log\ngit commit --amend --no-edit\n\n# Push\ngit push\n```\n\n### 5.6 Update Milestone Status\n\n1. Set milestone `Status: ACTIVE` → `Status: COMPLETE` in tasks.md\n2. Update milestone `Result Log:` with summary\n3. Update TodoWrite milestone todo to `completed`\n\n### 5.7 Continue to Next Milestone\n\n- Move to next PENDING milestone\n- Repeat steps 5.1-5.6\n\n---\n\n## 6. Plan Completion\n\n**When ALL milestones have Status: COMPLETE:**\n\n### 6.1 Create Final Git Tag\n\n```bash\nTASK_NAME=\"[task-name]\"\nSTART_TIMESTAMP=\"[from first provenance log entry]\"\nEND_TIMESTAMP=$(date -Iseconds)\nTOTAL_MILESTONES=\"[count from tasks.md]\"\n\ngit tag -a \"${TASK_NAME}-complete\" -m \"Completed AA-MA plan: $TASK_NAME\n\nTotal milestones: $TOTAL_MILESTONES\nStarted: $START_TIMESTAMP\nCompleted: $END_TIMESTAMP\n\nSee ${TASK_NAME}-provenance.log for full execution history.\"\n```\n\n### 6.2 Push Tag\n\n```bash\ngit push --tags\n```\n\n### 6.3 Final Provenance Entry\n\n```bash\nTIMESTAMP=$(date -Iseconds)\necho \"[$TIMESTAMP] PLAN COMPLETE — Tag: ${TASK_NAME}-complete — All milestones COMPLETE\" >> .claude\u002Fdev\u002Factive\u002F$TASK_NAME\u002F${TASK_NAME}-provenance.log\n\n# If a \u002Fgoal was bound at §2.5, capture the final verdict.\n# Query \u002Fgoal status (no args) and parse the evaluator's last reason.\n# Format: GOAL_FINAL — verdict=\u003CMET|UNMET|TURN_CAP_HIT|CLEARED> turns=\u003CN> reason='\u003Cshort>'\necho \"[$TIMESTAMP] GOAL_FINAL — verdict=[MET|UNMET|TURN_CAP_HIT|CLEARED|NOT_BOUND] turns=[N] reason='[short]'\" >> .claude\u002Fdev\u002Factive\u002F$TASK_NAME\u002F${TASK_NAME}-provenance.log\n\n# Commit provenance log\ngit add .claude\u002Fdev\u002Factive\u002F$TASK_NAME\u002F${TASK_NAME}-provenance.log\ngit commit -m \"docs($TASK_NAME): finalize provenance log\"\ngit push\n```\n\n### 6.4 Mark Plan Complete\n\nUpdate tasks.md:\n- Add header comment at top: `# Status: COMPLETE`\n- Add completion timestamp\n- Preserve all milestone\u002Ftask structure for reference\n\n---\n\n## 7. Report Completion\n\n**Success message**:\n```\n? AA-MA Plan Completed: [task-name]\n\nExecution Summary:\n- Total milestones: [count]\n- Total sub-tasks: [count]\n- Started: [start timestamp]\n- Completed: [end timestamp]\n- Duration: [calculated duration]\n\nGit:\n- Total commits: [count milestone commits]\n- Tag created: [task-name]-complete\n- All changes pushed to remote\n\nProvenance:\n- Full execution history: .claude\u002Fdev\u002Factive\u002F[task-name]\u002F[task-name]-provenance.log\n\nNext Steps:\n- Review implementation and verify all acceptance criteria met\n- Run end-to-end tests if not already executed\n- Archive completed plan: \u002Farchive-aa-ma [task-name]\n- Update project documentation if needed\n```\n\n---\n\n## 8. Error Handling\n\nSee [errors-and-operations.md](references\u002Ferrors-and-operations.md#8-error-handling) — validation failures, broken dependency chains, context limits, token use, rollback and best practices for a full run.\n\n**If a validation check fails at a milestone boundary, follow §8.1 in the reference first: set the milestone BLOCKED, add a remediation sub-task, alert the user and stop the run.**\n\n---\n\n**End of execute-aa-ma-full.md**\n","# AA-MA Full Plan Execution\n\nYou are executing the **complete AA-MA plan** from current position to final completion.\n\n**Scope**: All remaining milestones in tasks.md\n**Validation**: Strict at every milestone boundary\n**Git**: Auto-commit + provenance at each milestone, final tag on completion\n\n⚠️ **Use this command only for well-understood plans that don't need frequent human checkpoints.**\n\n---\n\n## 1. Locate Active Task & Pre-Execution Validation\n\nSee [pre-execution-validation.md](references\u002Fpre-execution-validation.md#1-locate-active-task--pre-execution-validation) — the Tier 1 existence check and the Tier 2 validator dispatch, with auto-recovery.\n\n---\n\n## 2. Full Context Injection\n\nSee [context-and-goal.md](references\u002Fcontext-and-goal.md#2-full-context-injection) — which AA-MA files to load for a full run, and the optional \u002Fgoal synthesis and bind for AFK runs.\n\n---\n\n## 2.5 Goal Synthesis & Bind (Optional, Recommended for AFK Runs)\n\nSee [context-and-goal.md](references\u002Fcontext-and-goal.md#25-goal-synthesis--bind-optional-recommended-for-afk-runs) — which AA-MA files to load for a full run, and the optional \u002Fgoal synthesis and bind for AFK runs.\n\n---\n\n## 3. Validate Loaded Context\n\n**Required checks**:\n- ✅ All 5 AA-MA files exist and are readable\n- ✅ REFERENCE contains key constants\n- ✅ TASKS has clear HTP structure with Status fields\n- ✅ At least one milestone has `Status: PENDING` or `Status: ACTIVE`\n- ✅ All milestones have explicit Acceptance Criteria defined\n- ✅ Dependencies between milestones are clearly specified\n\n**If validation fails**:\n```\nERROR: [Specific issue]\n- Missing files → \"AA-MA files incomplete. Run \u002Faa-ma-plan or check directory.\"\n- Malformed HTP → \"tasks.md lacks proper HTP format.\"\n- No pending milestones → \"All milestones complete.\"\n- Missing acceptance criteria → \"One or more milestones lack acceptance criteria.\"\n- Unclear dependencies → \"Milestone dependencies not specified.\"\n```\n\n---\n\n## 4. Parse HTP & Create TodoWrite Todos\n\n**Parse tasks.md for complete HTP structure**:\n- `## headers` = Milestones\n- `### headers` = Sub-tasks\n- Extract all metadata: `Status:`, `Dependencies:`, `Complexity:`, `Acceptance Criteria:`\n\n**Identify execution plan**:\n1. Count total milestones\n2. Count PENDING\u002FACTIVE milestones\n3. Count COMPLETE milestones\n4. Identify first PENDING milestone (starting point)\n5. Map dependency tree\n\n**Create TodoWrite todos for ALL remaining work**:\n- All PENDING\u002FACTIVE milestones\n- All sub-tasks within those milestones\n\n**Todo format**:\n```json\n{\n  \"content\": \"[Milestone\u002FTask title]\",\n  \"status\": \"pending|in_progress|completed\",\n  \"activeForm\": \"[Present continuous form]\"\n}\n```\n\n**Hierarchical organization**:\n```\n- [ ] Milestone 1: [title]\n  - [ ] Sub-task 1.1\n  - [ ] Sub-task 1.2\n- [ ] Milestone 2: [title]\n  - [ ] Sub-task 2.1\n  - [ ] Sub-task 2.2\n```\n\n---\n\n## 4.5 Quality Standards (Actionable Triggers)\n\nThese principles apply DURING execution. Each has a trigger condition and skip condition.\n\n- **KISS:** If implementation exceeds 200 lines for a single function or method, decompose before proceeding. Skip for generated code or data tables.\n- **DRY:** Before writing a new utility function, grep the codebase for existing implementations. If a match exists, reuse it. Section B.5 (Post-Milestone Simplification Review) catches misses post-hoc.\n- **SOLID:** If a class has >5 public methods, evaluate single-responsibility. If a module mixes data access + business logic + presentation, separate. Skip for scripts and one-off tools.\n- **SOC:** If a file grows beyond one clear responsibility, split. If a function handles both data transformation and side effects, separate.\n- **TDD:** If this milestone produces code, invoke `superpowers:test-driven-development`. Let the skill decide test strategy. Skip it only with a canonical `TDD-Waiver:` value (`engineering-standards.md` §2).\n- **12-Factor:** If the task involves service deployment, API servers, or containerized applications, reference 12-Factor principles for config (env vars not files), statelessness, and port binding. For ALL tasks with `.env` files, verify env-var-drift compliance (see `rules\u002Fenv-var-drift.md`).\n- **Context7 MCP:** Use for library docs and code generation. Retry once on failure, then fall back to WebSearch + official docs.\n- **WebSearch:** Fallback when Context7 fails. Also use proactively for unfamiliar APIs, recent library changes, or deployment patterns.\n\n### Pre-Execution Check (First Milestone Only)\n\nIf this is the **first milestone** and it touches 3+ files or unfamiliar code, invoke `Skill(system-mapping)` for the 5-point pre-flight check. Skip for subsequent milestones.\n\n---\n\n## 5. Execute Plan Sequentially\n\n**For each PENDING milestone (in order)**:\n\n### 5.1 Pre-Flight Checks\n\n1. **Update status**:\n   - Set milestone `Status: PENDING` → `Status: ACTIVE`\n   - Update TodoWrite milestone todo to `in_progress`\n\n2. **Dependency advisory — never blocks** (map Ticket 16): run the fence from\n   `\u002Fexecute-aa-ma-milestone` §5.1 step 3 (`aa_ma_deps advisory \"${TASKS_MD}\"`), show the\n   user any line it prints, log it to provenance.log, and continue.\n\n### 5.2 Execute All Sub-Tasks\n\n**For each sub-task in milestone**:\n1. Mark TodoWrite sub-task as `in_progress`\n1.5. **Mode Dispatch**: Parse `Mode:` from sub-task (inherit from milestone if absent, default HITL). HITL: display summary + criteria, `AskUserQuestion` [Proceed\u002FSkip\u002FAbort]. AFK: execute directly without pause.\n2. Execute sub-task (use agents, skills, Context7 MCP as needed; before editing existing code, run `Skill(impact-analysis)` on it)\n3. Verify result (guidance-based, trust your assessment)\n4. Update `Result Log:` in tasks.md\n5. Mark TodoWrite sub-task as `completed`\n\n### 5.3 Milestone Boundary Validation (STRICT)\n\n**REQUIRED validation before marking milestone COMPLETE:**\n\n#### A. Acceptance Criteria Verification\n- Explicitly verify EACH criterion met\n- Document verification method\u002Fresult for each\n- If ANY fails → set `Status: BLOCKED`, HALT, alert user\n\n#### B. Dependency Verification (for next milestone)\n- Advisory only: the next milestone's `Dependencies:` are reported by its own §5.1 step 2\n  when it becomes ACTIVE. Never halt on them.\n\n#### C. Impact Analysis Verification (REQUIRED)\n\n**Before completing milestone, verify no breaking changes introduced:**\n\n1. **Invoke the impact-analysis skill** (`Skill(impact-analysis)`)\n2. **Output consolidated impact analysis** for ALL files modified in this milestone\n3. **Verify no unresolved cascade effects**\n\n**Required output format** (consolidated for milestone):\n```\n? Impact Analysis: Milestone \"[Milestone Title]\"\n┌─ Files Modified: [N]\n│\n├─ [file_path]\n│  ├─ Upstream: [N] callers\n│  ├─ Contract: [YES\u002FNO]\n│  └─ Risk: [LOW\u002FMEDIUM\u002FHIGH]\n│\n└─ Overall Risk: [LOW\u002FMEDIUM\u002FHIGH]\n```\n\n**Validation rules**:\n- Overall Risk = LOW → Proceed to Test Execution\n- Overall Risk = MEDIUM → Document cascade updates, then proceed\n- Overall Risk = HIGH → HALT, present options (auto-fix, manual review, abort)\n\n**Never complete milestone with unresolved HIGH risk impacts.**\n\n#### D. Test Execution (if specified)\n- Run ALL listed tests\n- ALL must pass (zero failures)\n- Document results in Result Log\n- If failures → set `Status: BLOCKED`, HALT, alert user\n\n### 5.4 Finalization Protocol — MANDATORY\n\nBefore marking the milestone COMPLETE and creating git checkpoint, execute this 4-step finalization protocol at EACH milestone boundary. **No exceptions.**\n\n#### A. Integrity Check (Checklist Verification)\n\nDisplay acceptance criteria verification to the user:\n\n```\nAcceptance Criteria Verification:\n- ✓ [Criterion 1]: Confirmed - [brief evidence]\n- ✓ [Criterion 2]: Confirmed - [brief evidence]\n\nGate: [HARD|SOFT] — [Approval artifact found \u002F Convention-based]\nAll [X] criteria verified. Ready for finalization.\n```\n\n**Rules:**\n- Every acceptance criterion must be explicitly listed with evidence\n- If ANY criterion cannot be confirmed → HALT, do not proceed\n- If `Gate: HARD` and no approval artifact in context-log.md → HALT, request approval (see `\u002Fexecute-aa-ma-milestone` Section 7.1)\n- If `[task]-tests.yaml` exists, auto-run tests for this milestone (see `\u002Fexecute-aa-ma-milestone` Section 6.4)\n\n#### B. Documentation Auto-Update\n\nAutomatically update all 5 AA-MA files:\n\n| File | Auto-Update Action |\n|------|-------------------|\n| `tasks.md` | Mark milestone `Status: COMPLETE`, fill `Result Log:` |\n| `reference.md` | Add any new immutable facts discovered |\n| `context-log.md` | Append completion summary |\n| `provenance.log` | Append completion entry (after commit) |\n| `plan.md` | No change (historical record) |\n\n#### B.5. Post-Milestone Simplification Review\n\n**Skip if:** Only docs\u002Fconfig changed, diff \u003C 20 lines code, or `--skip-simplify` flag passed to `\u002Fexecute-aa-ma-full`.\n\nLaunch 3 parallel agents against milestone diff (reuse, quality, efficiency). CRITICAL findings require acknowledgment at approval gate. See `\u002Fexecute-aa-ma-milestone` Section 6.6 for full protocol.\n\n**Note:** For fully automated AFK runs, pass `--skip-simplify` to skip this review at every milestone and save tokens. The review is most valuable when running milestones interactively.\n\n**On failure:** Skip review, log to provenance. Do NOT block finalization.\n\n#### B.6. Post-Impl Adversarial Review (Phase 6.8 — NEW in v0.8.0)\n\nReference: ADR-0005 + `\u002Fexecute-aa-ma-milestone` Section 6.8 (full anatomy).\n\n**Skip if:** plan's `Created:` date is \u003C v0.8.0 release tag commit date (grandfathered), OR `AA_MA_AUDIT_BUDGET=off`, OR `AA_MA_HOOKS_DISABLE=1`.\n\nDelegate to `\u002Fexecute-aa-ma-milestone` Section 6.8 — same logic, same agents:\nthe `verify-impl` skill orchestrator dispatches up to 5 audit agents\n(code-reviewer, security-auditor, tdd-sequence-auditor,\ncontext7-evidence-auditor, future-proofing-auditor) per the milestone's\nplan-declared `Audit-Profile:` field. Aggregated verdict written to\n`[task]-impl-review.md`.\n\n**Verdicts:**\n- `PASS` \u002F `PASS_WITH_WARNINGS` → proceed to C (User Authorization)\n- `BLOCKED` (any CRITICAL accepted via override panel) → HALT this milestone, surface remediation to user. Do NOT auto-advance to next milestone. The user must fix the accepted CRITICAL, re-run `\u002Fexecute-aa-ma-milestone` (which re-runs §6.8), and only when §6.8 passes clean can `\u002Fexecute-aa-ma-full` resume.\n\n**Provenance:** §6.8 entry emitted by the milestone command itself. No duplicate entry from the full command — the delegation is logical, not physical.\n\n#### C. User Authorization (Approval Gate)\n\nUse AskUserQuestion to get explicit user approval:\n\n```\n? Finalization Review\n\nMilestone: [Title]\nAcceptance Criteria: [X\u002FX] verified ✓\n\nReady to mark COMPLETE?\n- Approve: Proceed with commit\n- Review First: Show detailed verification\n- Reject: Keep as ACTIVE\n```\n\n**Behavior:**\n- **Approve**: Proceed to git commit\n- **Review First**: Display full checklist, then re-prompt\n- **Reject**: HALT execution, keep `Status: ACTIVE`, ask user what needs fixing\n\n#### D. Transparent Status Change\n\nAfter approval, display minimal confirmation:\n\n```\n✅ Milestone marked COMPLETE: [Title]\n```\n\nThen proceed to git commit.\n\n### 5.5 Git Checkpoint Creation\n\n**If validation passes:**\n\n```bash\n# Stage changes\ngit add .\n\n# Get milestone info\nTASK_NAME=\"[task-name]\"\nMILESTONE_TITLE=\"[milestone title]\"\nMILESTONE_ID=\"[milestone-id]\"\nCOMPLEXITY=\"[complexity %]\"\n\n# Create structured commit\ngit commit -m \"feat($TASK_NAME): Complete $MILESTONE_TITLE\n\nMilestone: $MILESTONE_ID\nComplexity: $COMPLEXITY%\n\nAcceptance criteria verified:\n- [criterion 1]: ✓ [method]\n- [criterion 2]: ✓ [method]\n\nTasks completed:\n- [sub-task 1]\n- [sub-task 2]\n\nTests: [status]\n\n[AA-MA Plan] $TASK_NAME .claude\u002Fdev\u002Factive\u002F$TASK_NAME\"\n\n# Update provenance log\nCOMMIT_HASH=$(git rev-parse --short HEAD)\nTIMESTAMP=$(date -Iseconds)\necho \"[$TIMESTAMP] Commit $COMMIT_HASH — MilestoneID: $MILESTONE_ID — Status: COMPLETE\" >> .claude\u002Fdev\u002Factive\u002F$TASK_NAME\u002F${TASK_NAME}-provenance.log\n\n# Commit provenance update\ngit add .claude\u002Fdev\u002Factive\u002F$TASK_NAME\u002F${TASK_NAME}-provenance.log\ngit commit --amend --no-edit\n\n# Push\ngit push\n```\n\n### 5.6 Update Milestone Status\n\n1. Set milestone `Status: ACTIVE` → `Status: COMPLETE` in tasks.md\n2. Update milestone `Result Log:` with summary\n3. Update TodoWrite milestone todo to `completed`\n\n### 5.7 Continue to Next Milestone\n\n- Move to next PENDING milestone\n- Repeat steps 5.1-5.6\n\n---\n\n## 6. Plan Completion\n\n**When ALL milestones have Status: COMPLETE:**\n\n### 6.1 Create Final Git Tag\n\n```bash\nTASK_NAME=\"[task-name]\"\nSTART_TIMESTAMP=\"[from first provenance log entry]\"\nEND_TIMESTAMP=$(date -Iseconds)\nTOTAL_MILESTONES=\"[count from tasks.md]\"\n\ngit tag -a \"${TASK_NAME}-complete\" -m \"Completed AA-MA plan: $TASK_NAME\n\nTotal milestones: $TOTAL_MILESTONES\nStarted: $START_TIMESTAMP\nCompleted: $END_TIMESTAMP\n\nSee ${TASK_NAME}-provenance.log for full execution history.\"\n```\n\n### 6.2 Push Tag\n\n```bash\ngit push --tags\n```\n\n### 6.3 Final Provenance Entry\n\n```bash\nTIMESTAMP=$(date -Iseconds)\necho \"[$TIMESTAMP] PLAN COMPLETE — Tag: ${TASK_NAME}-complete — All milestones COMPLETE\" >> .claude\u002Fdev\u002Factive\u002F$TASK_NAME\u002F${TASK_NAME}-provenance.log\n\n# If a \u002Fgoal was bound at §2.5, capture the final verdict.\n# Query \u002Fgoal status (no args) and parse the evaluator's last reason.\n# Format: GOAL_FINAL — verdict=\u003CMET|UNMET|TURN_CAP_HIT|CLEARED> turns=\u003CN> reason='\u003Cshort>'\necho \"[$TIMESTAMP] GOAL_FINAL — verdict=[MET|UNMET|TURN_CAP_HIT|CLEARED|NOT_BOUND] turns=[N] reason='[short]'\" >> .claude\u002Fdev\u002Factive\u002F$TASK_NAME\u002F${TASK_NAME}-provenance.log\n\n# Commit provenance log\ngit add .claude\u002Fdev\u002Factive\u002F$TASK_NAME\u002F${TASK_NAME}-provenance.log\ngit commit -m \"docs($TASK_NAME): finalize provenance log\"\ngit push\n```\n\n### 6.4 Mark Plan Complete\n\nUpdate tasks.md:\n- Add header comment at top: `# Status: COMPLETE`\n- Add completion timestamp\n- Preserve all milestone\u002Ftask structure for reference\n\n---\n\n## 7. Report Completion\n\n**Success message**:\n```\n? AA-MA Plan Completed: [task-name]\n\nExecution Summary:\n- Total milestones: [count]\n- Total sub-tasks: [count]\n- Started: [start timestamp]\n- Completed: [end timestamp]\n- Duration: [calculated duration]\n\nGit:\n- Total commits: [count milestone commits]\n- Tag created: [task-name]-complete\n- All changes pushed to remote\n\nProvenance:\n- Full execution history: .claude\u002Fdev\u002Factive\u002F[task-name]\u002F[task-name]-provenance.log\n\nNext Steps:\n- Review implementation and verify all acceptance criteria met\n- Run end-to-end tests if not already executed\n- Archive completed plan: \u002Farchive-aa-ma [task-name]\n- Update project documentation if needed\n```\n\n---\n\n## 8. Error Handling\n\nSee [errors-and-operations.md](references\u002Ferrors-and-operations.md#8-error-handling) — validation failures, broken dependency chains, context limits, token use, rollback and best practices for a full run.\n\n**If a validation check fails at a milestone boundary, follow §8.1 in the reference first: set the milestone BLOCKED, add a remediation sub-task, alert the user and stop the run.**\n\n---\n\n**End of execute-aa-ma-full.md**\n",{"name":5,"description":7,"disable-model-invocation":36},true,"main","cd9a2962c9a7e0797fe440f7b1ba4ceccc16ae0d","claude-code\u002Fskills\u002Fexecute-aa-ma-full",1,"https:\u002F\u002Fgithub.com\u002Fsnewhouse\u002Faa-ma-forge","https:\u002F\u002Fgithub.com\u002Fsnewhouse","Data Scientist, Bioinformatician, Dad, Husband, MS warrior","2026-11-11 21:34:39",{"source":8,"skillName":5,"commands":46,"prompt":56,"zipAvailable":36},[47,50,53],{"tool":48,"command":49},"npx","npx skills add snewhouse\u002Faa-ma-forge --skill execute-aa-ma-full",{"tool":51,"command":52},"bunx","bunx skills add snewhouse\u002Faa-ma-forge --skill execute-aa-ma-full",{"tool":54,"command":55},"pnpm","pnpm dlx skills add snewhouse\u002Faa-ma-forge --skill execute-aa-ma-full","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fsnewhouse\u002Faa-ma-forge\u002Fblob\u002Fmain\u002Fclaude-code\u002Fskills\u002Fexecute-aa-ma-full\u002FSKILL.md",[58,64,68],{"path":59,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":62,"sourceHtmlUrl":63},"references\u002Fcontext-and-goal.md","REFERENCE","参考文档",4621,"https:\u002F\u002Fgithub.com\u002Fsnewhouse\u002Faa-ma-forge\u002Fblob\u002Fmain\u002Fclaude-code\u002Fskills\u002Fexecute-aa-ma-full\u002Freferences\u002Fcontext-and-goal.md",{"path":65,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":66,"sourceHtmlUrl":67},"references\u002Ferrors-and-operations.md",5027,"https:\u002F\u002Fgithub.com\u002Fsnewhouse\u002Faa-ma-forge\u002Fblob\u002Fmain\u002Fclaude-code\u002Fskills\u002Fexecute-aa-ma-full\u002Freferences\u002Ferrors-and-operations.md",{"path":69,"resourceType":60,"resourceTypeDesc":61,"sizeBytes":70,"sourceHtmlUrl":71},"references\u002Fpre-execution-validation.md",3159,"https:\u002F\u002Fgithub.com\u002Fsnewhouse\u002Faa-ma-forge\u002Fblob\u002Fmain\u002Fclaude-code\u002Fskills\u002Fexecute-aa-ma-full\u002Freferences\u002Fpre-execution-validation.md",[73,97,125,144,169,199],{"skillNo":74,"name":75,"nameZh":6,"description":76,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":77,"categories":78,"keywords":84,"publishedAt":31},"01M6MZ588SD5DDH59H18PRNWEV","deploy-helper","You can use this to help with deploy stuff, deploying, deployments, releases, shipping, pushing to prod, going live, rollouts.","https:\u002F\u002Fgithub.com\u002Fsnewhouse\u002Faa-ma-forge\u002Fblob\u002Fmain\u002Ftests\u002Ffixtures\u002Fevals\u002Fskill-draft\u002Fdraft\u002FSKILL.md",[79],{"categoryNo":80,"categoryName":81,"taxonomyType":82,"taxonomyTypeDesc":83},"01JBSEEDCAT000000000000011","运维与部署","SCENARIO","使用场景",[85,86,87,88,89,28,90,91,92,93,94,95,96],"deploy","deploy helper","helper","you","live","can","rollouts","shipping","use","deployments","releases","help",{"skillNo":98,"name":99,"nameZh":6,"description":100,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":101,"licenseFileUrl":102,"sourceHtmlUrl":103,"categories":104,"keywords":116,"publishedAt":31},"01M6MZ54NEKRYXGMJ1KJWYMKM7","writing-for-agents","Writing documents for agents. Use when creating or editing skills, or modifying AGENTS.md or CLAUDE.md.","MIT","https:\u002F\u002Fgithub.com\u002Fsnewhouse\u002Faa-ma-forge\u002Fblob\u002Fmain\u002Fclaude-code\u002Fskills\u002Fwriting-for-agents\u002FLICENSE","https:\u002F\u002Fgithub.com\u002Fsnewhouse\u002Faa-ma-forge\u002Fblob\u002Fmain\u002Fclaude-code\u002Fskills\u002Fwriting-for-agents\u002FSKILL.md",[105,108,111],{"categoryNo":106,"categoryName":107,"taxonomyType":82,"taxonomyTypeDesc":83},"01JBSEEDCAT000000000000018","写作与文案",{"categoryNo":109,"categoryName":110,"taxonomyType":82,"taxonomyTypeDesc":83},"01JBSEEDCAT000000000000023","Skill与智能体开发",{"categoryNo":112,"categoryName":113,"taxonomyType":114,"taxonomyTypeDesc":115},"01JBSEEDCAT000000000000036","内容创作者","PROFESSION","适用职业",[117,118,119,120,93,121,122,123,124],"writing","writing for agents","for","line","instructions","read","editing","documents",{"skillNo":126,"name":127,"nameZh":6,"description":128,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":129,"categories":130,"keywords":131,"publishedAt":31},"01M6MZ500T0FH1G8TBSQV7G5G2","verify-plan","Slash entry for plan verification: runs plan-verification on an existing AA-MA plan, writes [task]-verification.md, and with --iterate loops until the plan converges. Use when the user types \u002Fverify-plan or wants the verification report saved.","https:\u002F\u002Fgithub.com\u002Fsnewhouse\u002Faa-ma-forge\u002Fblob\u002Fmain\u002Fclaude-code\u002Fskills\u002Fverify-plan\u002FSKILL.md",[],[132,23,133,134,135,93,24,28,136,137,138,139,140,141,142,119,143],"verify","verify plan","loops","task","verification","report","types","plan verification","entry","iterate","Existing","saved",{"skillNo":145,"name":146,"nameZh":6,"description":147,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":148,"categories":149,"keywords":150,"publishedAt":31},"01M6MZ4V9ST4K54NT1SRBGCE7P","verify-impl","Post-impl adversarial review symmetric to \u002Fverify-plan. Dispatches up to 5 parallel audit agents (code-reviewer, security-auditor, tdd-sequence-auditor, context7-evidence-auditor, future-proofing-auditor) based on the milestone's plan-declared Audit-Profile. CRITICAL findings surface via AskUserQuestion accept\u002Fdispute\u002Fdefer panel before §7.3 user authorization. Introduced in v0.8.0 per ADR-0005. Invoked by Phase 6.8 of \u002Fexecute-aa-ma-milestone.","https:\u002F\u002Fgithub.com\u002Fsnewhouse\u002Faa-ma-forge\u002Fblob\u002Fmain\u002Fclaude-code\u002Fskills\u002Fverify-impl\u002FSKILL.md",[],[151,152,132,153,154,155,156,157,158,159,160,161,162,163,164,23,165,166,167,168],"impl","verify impl","adversarial","post","panel","via","critical","review","context7","up","evidence","reviewer","parallel","adr","audit","code","tdd","milestone",{"skillNo":170,"name":171,"nameZh":6,"description":172,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":173,"categories":174,"keywords":179,"publishedAt":31},"01M6MZ4K7CC9PMJ1KEY3F53RS1","understand-codebase","Onboard to a new, inherited, or shared codebase. Read it, understand it, map it, and learn its conventions, versioning, tests, tech stack, rules (AGENTS.md \u002F CLAUDE.md \u002F .cursorrules \u002F etc.), build\u002Frun\u002FCI, integrations, security posture, and repo health — then produce an honest pros\u002Fcons verdict, a \"contribute safely\" playbook, and an \"add a feature\" playbook, written to ONBOARDING.md at the repo root plus a .claude\u002Fonboarding\u002F set of deep-dives — and optionally author an AGENTS.md if one is missing, or review and propose improvements to an existing one. Tiered (Quick \u002F Standard \u002F Deep). Reuses codemem (index + diagrams), gsd-map-codebase, system-mapping, code-intelligence, impact-analysis rather than re-implementing them; Deep tier runs a TeamCreate agent-team. Keywords: new codebase, shared codebase, inherited code, onboard, understand this repo, how do I contribute, how do I add a feature, ramp up, get oriented, ONBOARDING.md, AGENTS.md, codebase walkthrough, joining a project.","https:\u002F\u002Fgithub.com\u002Fsnewhouse\u002Faa-ma-forge\u002Fblob\u002Fmain\u002Fclaude-code\u002Fskills\u002Funderstand-codebase\u002FSKILL.md",[175,176],{"categoryNo":109,"categoryName":110,"taxonomyType":82,"taxonomyTypeDesc":83},{"categoryNo":177,"categoryName":178,"taxonomyType":114,"taxonomyTypeDesc":115},"01JBSEEDCAT000000000000038","人力资源",[180,181,182,160,183,184,185,186,187,188,189,190,191,192,193,194,195,196,197,198],"understand codebase","codebase","understand","walkthrough","analysis","SET","ci","learn","diagrams","repo health","implementing","re","quick","impact","get","project","deep","shared","map",{"skillNo":200,"name":201,"nameZh":6,"description":202,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":203,"categories":204,"keywords":205,"publishedAt":31},"01M6MZ4D1ZC8MTW064QXY6WC76","token-compression","Output token compression for AA-MA workflows. Reduces response tokens by 40-75% while preserving full technical accuracy. Three intensity levels mapped to execution modes: lite (HITL), full (general), ultra (AFK). Auto-activates during AA-MA execution based on task Mode field. Manual activation via \u002Fcompress lite|full|ultra.","https:\u002F\u002Fgithub.com\u002Fsnewhouse\u002Faa-ma-forge\u002Fblob\u002Fmain\u002Fclaude-code\u002Fskills\u002Ftoken-compression\u002FSKILL.md",[],[206,207,208,209,210,211,212,119,213,214,156,215,216,217,218,135,219,220,221,222],"compression","token compression","token","workflows","output","modes","accuracy","mode","manual","intensity","tokens","hitl","three","general","auto","technical","execution"]