@thierrynakoa/fire-flow 10.0.0
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.
- package/.claude-plugin/plugin.json +64 -0
- package/ARCHITECTURE-DIAGRAM.md +440 -0
- package/COMMAND-REFERENCE.md +172 -0
- package/DOMINION-FLOW-OVERVIEW.md +421 -0
- package/LICENSE +21 -0
- package/QUICK-START.md +351 -0
- package/README.md +398 -0
- package/TROUBLESHOOTING.md +264 -0
- package/agents/fire-codebase-mapper.md +484 -0
- package/agents/fire-debugger.md +535 -0
- package/agents/fire-executor.md +949 -0
- package/agents/fire-fact-checker.md +276 -0
- package/agents/fire-learncoding-explainer.md +237 -0
- package/agents/fire-learncoding-walker.md +147 -0
- package/agents/fire-planner.md +675 -0
- package/agents/fire-project-researcher.md +155 -0
- package/agents/fire-research-synthesizer.md +166 -0
- package/agents/fire-researcher.md +723 -0
- package/agents/fire-reviewer.md +499 -0
- package/agents/fire-roadmapper.md +203 -0
- package/agents/fire-verifier.md +880 -0
- package/bin/cli.js +208 -0
- package/commands/fire-0-orient.md +476 -0
- package/commands/fire-1-new.md +281 -0
- package/commands/fire-1a-discuss.md +455 -0
- package/commands/fire-2-plan.md +527 -0
- package/commands/fire-3-execute.md +1303 -0
- package/commands/fire-4-verify.md +845 -0
- package/commands/fire-5-handoff.md +515 -0
- package/commands/fire-6-resume.md +501 -0
- package/commands/fire-7-review.md +409 -0
- package/commands/fire-add-new-skill.md +598 -0
- package/commands/fire-analytics.md +499 -0
- package/commands/fire-assumptions.md +78 -0
- package/commands/fire-autonomous.md +528 -0
- package/commands/fire-brainstorm.md +413 -0
- package/commands/fire-complete-milestone.md +270 -0
- package/commands/fire-dashboard.md +375 -0
- package/commands/fire-debug.md +663 -0
- package/commands/fire-discover.md +616 -0
- package/commands/fire-double-check.md +460 -0
- package/commands/fire-execute-plan.md +182 -0
- package/commands/fire-learncoding.md +242 -0
- package/commands/fire-loop-resume.md +272 -0
- package/commands/fire-loop-stop.md +198 -0
- package/commands/fire-loop.md +1168 -0
- package/commands/fire-map-codebase.md +313 -0
- package/commands/fire-new-milestone.md +356 -0
- package/commands/fire-reflect.md +235 -0
- package/commands/fire-research.md +246 -0
- package/commands/fire-search.md +330 -0
- package/commands/fire-security-audit-repo.md +293 -0
- package/commands/fire-security-scan.md +484 -0
- package/commands/fire-session-summary.md +252 -0
- package/commands/fire-skills-diff.md +506 -0
- package/commands/fire-skills-history.md +388 -0
- package/commands/fire-skills-rollback.md +408 -0
- package/commands/fire-skills-sync.md +470 -0
- package/commands/fire-test.md +520 -0
- package/commands/fire-todos.md +335 -0
- package/commands/fire-transition.md +186 -0
- package/commands/fire-update.md +312 -0
- package/commands/fire-verify-uat.md +146 -0
- package/commands/fire-vuln-scan.md +493 -0
- package/hooks/hooks.json +16 -0
- package/hooks/run-hook.cmd +69 -0
- package/hooks/run-hook.sh +8 -0
- package/hooks/run-session-end.cmd +49 -0
- package/hooks/run-session-end.sh +7 -0
- package/hooks/session-end.sh +90 -0
- package/hooks/session-start.sh +111 -0
- package/package.json +52 -0
- package/plugin.json +7 -0
- package/references/auto-skill-extraction.md +136 -0
- package/references/behavioral-directives.md +365 -0
- package/references/blocker-tracking.md +155 -0
- package/references/checkpoints.md +165 -0
- package/references/circuit-breaker.md +410 -0
- package/references/context-engineering.md +587 -0
- package/references/decision-time-guidance.md +289 -0
- package/references/error-classification.md +326 -0
- package/references/execution-mode-intelligence.md +242 -0
- package/references/git-integration.md +217 -0
- package/references/honesty-protocols.md +304 -0
- package/references/integration-architecture.md +470 -0
- package/references/issue-to-pr-pipeline.md +150 -0
- package/references/metrics-and-trends.md +234 -0
- package/references/playwright-e2e-testing.md +326 -0
- package/references/questioning.md +125 -0
- package/references/research-improvements.md +110 -0
- package/references/skills-usage-guide.md +429 -0
- package/references/tdd.md +131 -0
- package/references/testing-enforcement.md +192 -0
- package/references/ui-brand.md +383 -0
- package/references/validation-checklist.md +456 -0
- package/references/verification-patterns.md +187 -0
- package/references/warrior-principles.md +173 -0
- package/skills-library/SKILLS-INDEX.md +588 -0
- package/skills-library/_general/frontend/html-visual-reports.md +292 -0
- package/skills-library/_general/methodology/debug-swarm-researcher-escape-hatch.md +240 -0
- package/skills-library/_general/methodology/learncoding-agentic-pattern.md +114 -0
- package/skills-library/_general/methodology/shell-autonomous-loop-fixplan.md +238 -0
- package/skills-library/basics/api-rest-basics.md +162 -0
- package/skills-library/basics/env-variables.md +96 -0
- package/skills-library/basics/error-handling-basics.md +125 -0
- package/skills-library/basics/git-commit-conventions.md +106 -0
- package/skills-library/basics/readme-template.md +108 -0
- package/skills-library/common-tasks/async-await-patterns.md +157 -0
- package/skills-library/common-tasks/auth-jwt-basics.md +164 -0
- package/skills-library/common-tasks/database-schema-design.md +166 -0
- package/skills-library/common-tasks/file-upload-basics.md +166 -0
- package/skills-library/common-tasks/form-validation.md +159 -0
- package/skills-library/debugging/FAILURE_TAXONOMY_CLASSIFICATION.md +117 -0
- package/skills-library/debugging/THREE_AGENT_HYPOTHESIS_DEBUGGING.md +86 -0
- package/skills-library/methodology/BREATH_BASED_PARALLEL_EXECUTION.md +678 -0
- package/skills-library/methodology/CONFIDENCE_GATED_EXECUTION.md +243 -0
- package/skills-library/methodology/EVIDENCE_BASED_VALIDATION.md +308 -0
- package/skills-library/methodology/MULTI_PERSPECTIVE_CODE_REVIEW.md +330 -0
- package/skills-library/methodology/PATH_VERIFICATION_GATE.md +211 -0
- package/skills-library/methodology/REFLEXION_MEMORY_PATTERN.md +183 -0
- package/skills-library/methodology/RESEARCH_BACKED_WORKFLOW_UPGRADE.md +263 -0
- package/skills-library/methodology/SABBATH_REST_PATTERN.md +267 -0
- package/skills-library/methodology/STONE_AND_SCAFFOLD.md +220 -0
- package/skills-library/performance/cache-augmented-generation.md +172 -0
- package/skills-library/quality-safety/debugging-steps.md +147 -0
- package/skills-library/quality-safety/deployment-checklist.md +155 -0
- package/skills-library/quality-safety/security-checklist.md +204 -0
- package/skills-library/quality-safety/testing-basics.md +180 -0
- package/skills-library/security/agent-security-scanner.md +445 -0
- package/skills-library/specialists/api-architecture/api-designer.md +49 -0
- package/skills-library/specialists/api-architecture/graphql-architect.md +49 -0
- package/skills-library/specialists/api-architecture/mcp-developer.md +51 -0
- package/skills-library/specialists/api-architecture/microservices-architect.md +50 -0
- package/skills-library/specialists/api-architecture/websocket-engineer.md +48 -0
- package/skills-library/specialists/backend/django-expert.md +52 -0
- package/skills-library/specialists/backend/fastapi-expert.md +52 -0
- package/skills-library/specialists/backend/laravel-specialist.md +52 -0
- package/skills-library/specialists/backend/nestjs-expert.md +51 -0
- package/skills-library/specialists/backend/rails-expert.md +53 -0
- package/skills-library/specialists/backend/spring-boot-engineer.md +56 -0
- package/skills-library/specialists/data-ml/fine-tuning-expert.md +48 -0
- package/skills-library/specialists/data-ml/ml-pipeline.md +47 -0
- package/skills-library/specialists/data-ml/pandas-pro.md +47 -0
- package/skills-library/specialists/data-ml/rag-architect.md +51 -0
- package/skills-library/specialists/data-ml/spark-engineer.md +47 -0
- package/skills-library/specialists/frontend/angular-architect.md +52 -0
- package/skills-library/specialists/frontend/flutter-expert.md +51 -0
- package/skills-library/specialists/frontend/nextjs-developer.md +54 -0
- package/skills-library/specialists/frontend/react-native-expert.md +50 -0
- package/skills-library/specialists/frontend/vue-expert.md +51 -0
- package/skills-library/specialists/infrastructure/chaos-engineer.md +74 -0
- package/skills-library/specialists/infrastructure/cloud-architect.md +70 -0
- package/skills-library/specialists/infrastructure/database-optimizer.md +64 -0
- package/skills-library/specialists/infrastructure/devops-engineer.md +70 -0
- package/skills-library/specialists/infrastructure/kubernetes-specialist.md +52 -0
- package/skills-library/specialists/infrastructure/monitoring-expert.md +70 -0
- package/skills-library/specialists/infrastructure/sre-engineer.md +70 -0
- package/skills-library/specialists/infrastructure/terraform-engineer.md +51 -0
- package/skills-library/specialists/languages/cpp-pro.md +74 -0
- package/skills-library/specialists/languages/csharp-developer.md +69 -0
- package/skills-library/specialists/languages/dotnet-core-expert.md +54 -0
- package/skills-library/specialists/languages/golang-pro.md +51 -0
- package/skills-library/specialists/languages/java-architect.md +49 -0
- package/skills-library/specialists/languages/javascript-pro.md +68 -0
- package/skills-library/specialists/languages/kotlin-specialist.md +68 -0
- package/skills-library/specialists/languages/php-pro.md +49 -0
- package/skills-library/specialists/languages/python-pro.md +52 -0
- package/skills-library/specialists/languages/react-expert.md +51 -0
- package/skills-library/specialists/languages/rust-engineer.md +50 -0
- package/skills-library/specialists/languages/sql-pro.md +56 -0
- package/skills-library/specialists/languages/swift-expert.md +69 -0
- package/skills-library/specialists/languages/typescript-pro.md +51 -0
- package/skills-library/specialists/platform/atlassian-mcp.md +52 -0
- package/skills-library/specialists/platform/embedded-systems.md +53 -0
- package/skills-library/specialists/platform/game-developer.md +53 -0
- package/skills-library/specialists/platform/salesforce-developer.md +53 -0
- package/skills-library/specialists/platform/shopify-expert.md +49 -0
- package/skills-library/specialists/platform/wordpress-pro.md +49 -0
- package/skills-library/specialists/quality/code-documenter.md +51 -0
- package/skills-library/specialists/quality/code-reviewer.md +67 -0
- package/skills-library/specialists/quality/debugging-wizard.md +51 -0
- package/skills-library/specialists/quality/fullstack-guardian.md +51 -0
- package/skills-library/specialists/quality/legacy-modernizer.md +50 -0
- package/skills-library/specialists/quality/playwright-expert.md +65 -0
- package/skills-library/specialists/quality/spec-miner.md +56 -0
- package/skills-library/specialists/quality/test-master.md +65 -0
- package/skills-library/specialists/security/secure-code-guardian.md +55 -0
- package/skills-library/specialists/security/security-reviewer.md +53 -0
- package/skills-library/specialists/workflow/architecture-designer.md +53 -0
- package/skills-library/specialists/workflow/cli-developer.md +70 -0
- package/skills-library/specialists/workflow/feature-forge.md +65 -0
- package/skills-library/specialists/workflow/prompt-engineer.md +54 -0
- package/skills-library/specialists/workflow/the-fool.md +62 -0
- package/templates/ASSUMPTIONS.md +125 -0
- package/templates/BLOCKERS.md +73 -0
- package/templates/DECISION_LOG.md +116 -0
- package/templates/UAT.md +96 -0
- package/templates/blueprint.md +94 -0
- package/templates/brainstorm.md +185 -0
- package/templates/conscience.md +92 -0
- package/templates/fire-handoff.md +159 -0
- package/templates/metrics.md +67 -0
- package/templates/phase-prompt.md +142 -0
- package/templates/record.md +131 -0
- package/templates/review-report.md +117 -0
- package/templates/skills-index.md +157 -0
- package/templates/verification.md +149 -0
- package/templates/vision.md +79 -0
- package/validation-config.yml +793 -0
- package/version.json +7 -0
- package/workflows/execute-phase.md +732 -0
- package/workflows/handoff-session.md +678 -0
- package/workflows/new-project.md +578 -0
- package/workflows/plan-phase.md +592 -0
- package/workflows/verify-phase.md +874 -0
|
@@ -0,0 +1,528 @@
|
|
|
1
|
+
---
|
|
2
|
+
description: Full autopilot — plan, execute, verify all phases autonomously after PRD is complete
|
|
3
|
+
argument-hint: "[--from-phase N] [--to-phase N] [--max-phase-attempts 3]"
|
|
4
|
+
---
|
|
5
|
+
|
|
6
|
+
# /fire-autonomous
|
|
7
|
+
|
|
8
|
+
> Full autopilot after PRD is complete — plan, execute, verify, advance through all phases independently
|
|
9
|
+
|
|
10
|
+
---
|
|
11
|
+
|
|
12
|
+
## Purpose
|
|
13
|
+
|
|
14
|
+
After discussions and PRD are complete (`/fire-1a-discuss`), this command takes over the entire build pipeline. Claude plans each phase, executes it, verifies the work (verifier + reviewer), fixes any issues, and advances to the next phase — all without human checkpoints.
|
|
15
|
+
|
|
16
|
+
**Design principle (Boris Cherny):** "You don't trust; you instrument." All verification stays active. Verdicts auto-route to fix cycles instead of pausing for human input. The human reviews the finished product, not intermediate steps.
|
|
17
|
+
|
|
18
|
+
**Safety:** Path verification, HAC, circuit breaker, fire-verifier, and fire-reviewer all remain active. Autonomous mode removes human pauses, not safety gates.
|
|
19
|
+
|
|
20
|
+
---
|
|
21
|
+
|
|
22
|
+
## Arguments
|
|
23
|
+
|
|
24
|
+
```yaml
|
|
25
|
+
arguments:
|
|
26
|
+
--from-phase:
|
|
27
|
+
type: integer
|
|
28
|
+
default: "next incomplete phase from CONSCIENCE.md"
|
|
29
|
+
description: "Start from this phase number"
|
|
30
|
+
|
|
31
|
+
--to-phase:
|
|
32
|
+
type: integer
|
|
33
|
+
default: "last phase in VISION.md"
|
|
34
|
+
description: "Stop after completing this phase"
|
|
35
|
+
|
|
36
|
+
--max-phase-attempts:
|
|
37
|
+
type: integer
|
|
38
|
+
default: 3
|
|
39
|
+
description: "Max plan-execute-verify cycles per phase before escalating"
|
|
40
|
+
|
|
41
|
+
--max-iterations:
|
|
42
|
+
type: integer
|
|
43
|
+
default: 50
|
|
44
|
+
description: "Max iterations per execution loop (passed to fire-3-execute)"
|
|
45
|
+
|
|
46
|
+
--dry-run:
|
|
47
|
+
type: boolean
|
|
48
|
+
default: false
|
|
49
|
+
description: "Show what would be done without executing"
|
|
50
|
+
```
|
|
51
|
+
|
|
52
|
+
---
|
|
53
|
+
|
|
54
|
+
## Process
|
|
55
|
+
|
|
56
|
+
### Step 1: Validate Prerequisites
|
|
57
|
+
|
|
58
|
+
```
|
|
59
|
+
Read .planning/VISION.md → extract all phases
|
|
60
|
+
Read .planning/CONSCIENCE.md → get current status, find next incomplete phase
|
|
61
|
+
|
|
62
|
+
IF --from-phase not specified:
|
|
63
|
+
from_phase = first phase with status != complete in CONSCIENCE.md
|
|
64
|
+
|
|
65
|
+
IF --to-phase not specified:
|
|
66
|
+
to_phase = last phase in VISION.md
|
|
67
|
+
|
|
68
|
+
Validate:
|
|
69
|
+
1. .planning/ directory exists
|
|
70
|
+
2. VISION.md has phases in range [from_phase, to_phase]
|
|
71
|
+
3. For each phase in range: MEMORY.md exists (PRD/discussion complete)
|
|
72
|
+
4. Path Verification Gate (MANDATORY — working directory matches project)
|
|
73
|
+
|
|
74
|
+
IF any phase lacks MEMORY.md:
|
|
75
|
+
Display:
|
|
76
|
+
"Phase {N} has no MEMORY.md — discussions not complete.
|
|
77
|
+
Run /fire-1a-discuss {N} first.
|
|
78
|
+
Autonomous mode requires all target phases to have context."
|
|
79
|
+
STOP
|
|
80
|
+
```
|
|
81
|
+
|
|
82
|
+
### Step 0.5: Path Verification Gate (v5.0 — MANDATORY)
|
|
83
|
+
|
|
84
|
+
```
|
|
85
|
+
Same as all Dominion Flow commands:
|
|
86
|
+
1. Verify working directory matches expected project path
|
|
87
|
+
2. Check for cross-project contamination risk
|
|
88
|
+
3. Validate subagent path injection safety
|
|
89
|
+
|
|
90
|
+
This gate is NEVER disabled, not even in autonomous mode.
|
|
91
|
+
```
|
|
92
|
+
|
|
93
|
+
### Step 2: Display Autonomous Mode Banner
|
|
94
|
+
|
|
95
|
+
```
|
|
96
|
+
+--------------------------------------------------------------+
|
|
97
|
+
| AUTONOMOUS MODE ACTIVATED |
|
|
98
|
+
+--------------------------------------------------------------+
|
|
99
|
+
| |
|
|
100
|
+
| Project: {project name from VISION.md} |
|
|
101
|
+
| Phases: {from_phase} to {to_phase} ({count} phases) |
|
|
102
|
+
| Max attempts per phase: {max_phase_attempts} |
|
|
103
|
+
| |
|
|
104
|
+
| Active Safety: |
|
|
105
|
+
| Path verification — MANDATORY (cannot disable) |
|
|
106
|
+
| HAC enforcement — Active (confidence 5/5 rules) |
|
|
107
|
+
| Circuit breaker — Active (stall/spin/degrade detect) |
|
|
108
|
+
| fire-verifier — 70-point WARRIOR validation |
|
|
109
|
+
| fire-reviewer — 15-persona code review |
|
|
110
|
+
| |
|
|
111
|
+
| Human touchpoints: NONE until completion or blocker |
|
|
112
|
+
| |
|
|
113
|
+
| To interrupt: Ctrl+C or /fire-loop-stop |
|
|
114
|
+
| |
|
|
115
|
+
+--------------------------------------------------------------+
|
|
116
|
+
```
|
|
117
|
+
|
|
118
|
+
IF `--dry-run`:
|
|
119
|
+
```
|
|
120
|
+
DRY RUN — would execute:
|
|
121
|
+
|
|
122
|
+
Phase {N}: {name}
|
|
123
|
+
1. /fire-2-plan {N} --skip-checker
|
|
124
|
+
2. /fire-3-execute {N} --auto-continue --autonomous
|
|
125
|
+
3. Read verification + review results
|
|
126
|
+
4. Auto-fix if needed (up to {max_phase_attempts} attempts)
|
|
127
|
+
|
|
128
|
+
Phase {N+1}: {name}
|
|
129
|
+
...
|
|
130
|
+
|
|
131
|
+
No changes made. Remove --dry-run to execute.
|
|
132
|
+
```
|
|
133
|
+
STOP
|
|
134
|
+
|
|
135
|
+
### Step 3: Phase Loop
|
|
136
|
+
|
|
137
|
+
```
|
|
138
|
+
autonomous_log = []
|
|
139
|
+
phases_completed = 0
|
|
140
|
+
start_time = now()
|
|
141
|
+
|
|
142
|
+
FOR phase_number in range(from_phase, to_phase + 1):
|
|
143
|
+
|
|
144
|
+
phase_name = VISION.md phase name
|
|
145
|
+
attempt = 0
|
|
146
|
+
phase_complete = false
|
|
147
|
+
|
|
148
|
+
Display:
|
|
149
|
+
"━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━"
|
|
150
|
+
" AUTONOMOUS: Phase {phase_number} — {phase_name}"
|
|
151
|
+
"━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━"
|
|
152
|
+
|
|
153
|
+
WHILE attempt < max_phase_attempts AND NOT phase_complete:
|
|
154
|
+
attempt += 1
|
|
155
|
+
|
|
156
|
+
Display: " Attempt {attempt}/{max_phase_attempts}..."
|
|
157
|
+
|
|
158
|
+
// ──────────────────────────────────────────────────────
|
|
159
|
+
// Step 3.1: AUTO-PLAN
|
|
160
|
+
// ──────────────────────────────────────────────────────
|
|
161
|
+
|
|
162
|
+
phase_dir = .planning/phases/{phase_number}-{phase_name}/
|
|
163
|
+
plan_files = glob("{phase_dir}/*.BLUEPRINT.md")
|
|
164
|
+
|
|
165
|
+
IF plan_files is empty:
|
|
166
|
+
Display: " Planning phase {phase_number}..."
|
|
167
|
+
|
|
168
|
+
Run /fire-2-plan {phase_number} --skip-checker
|
|
169
|
+
// skip-checker: plan issues surface during verification
|
|
170
|
+
// auto-replan on failure is more effective than blocking on warnings
|
|
171
|
+
|
|
172
|
+
ELSE:
|
|
173
|
+
Display: " Plans exist ({count} plans). Executing..."
|
|
174
|
+
|
|
175
|
+
// ──────────────────────────────────────────────────────
|
|
176
|
+
// Step 3.2: AUTO-EXECUTE
|
|
177
|
+
// ──────────────────────────────────────────────────────
|
|
178
|
+
|
|
179
|
+
Display: " Executing phase {phase_number}..."
|
|
180
|
+
|
|
181
|
+
Run /fire-3-execute {phase_number} --auto-continue --autonomous
|
|
182
|
+
// --auto-continue: no breath-boundary interrupts
|
|
183
|
+
// --autonomous: merge gate auto-routes verdicts (v9.0)
|
|
184
|
+
|
|
185
|
+
// ──────────────────────────────────────────────────────
|
|
186
|
+
// Step 3.3: AUTO-VERIFY (read results)
|
|
187
|
+
// ──────────────────────────────────────────────────────
|
|
188
|
+
|
|
189
|
+
verification_file = {phase_dir}/{N}-VERIFICATION.md
|
|
190
|
+
review_file = {phase_dir}/{N}-REVIEW.md
|
|
191
|
+
|
|
192
|
+
Read verification_file → extract warrior_score, verifier_verdict
|
|
193
|
+
Read review_file → extract reviewer_verdict, finding_counts
|
|
194
|
+
|
|
195
|
+
// ──────────────────────────────────────────────────────
|
|
196
|
+
// Step 3.4: Evaluate Results
|
|
197
|
+
// ──────────────────────────────────────────────────────
|
|
198
|
+
|
|
199
|
+
IF verifier_verdict == "APPROVED" AND reviewer_verdict == "APPROVE":
|
|
200
|
+
phase_complete = true
|
|
201
|
+
phase_result = "CLEAN PASS"
|
|
202
|
+
|
|
203
|
+
Display: " Phase {N}: CLEAN PASS (verifier {score}/70, reviewer APPROVE)"
|
|
204
|
+
|
|
205
|
+
ELIF verifier_verdict in ["APPROVED", "CONDITIONAL"] AND reviewer_verdict in ["APPROVE", "APPROVE WITH FIXES"]:
|
|
206
|
+
phase_complete = true
|
|
207
|
+
phase_result = "PASSED WITH NOTES"
|
|
208
|
+
|
|
209
|
+
Display: " Phase {N}: PASSED WITH NOTES"
|
|
210
|
+
Display: " Verifier: {verifier_verdict} ({score}/70)"
|
|
211
|
+
Display: " Reviewer: {reviewer_verdict} ({finding_counts})"
|
|
212
|
+
Display: " Non-critical gaps logged to autonomous-log.md"
|
|
213
|
+
|
|
214
|
+
ELSE:
|
|
215
|
+
phase_result = "NEEDS FIX"
|
|
216
|
+
|
|
217
|
+
Display: " Phase {N}: NEEDS FIX (attempt {attempt})"
|
|
218
|
+
Display: " Verifier: {verifier_verdict}"
|
|
219
|
+
Display: " Reviewer: {reviewer_verdict}"
|
|
220
|
+
|
|
221
|
+
IF attempt < max_phase_attempts:
|
|
222
|
+
Display: " Re-planning for gaps..."
|
|
223
|
+
Run /fire-2-plan {phase_number} --gaps
|
|
224
|
+
// Loop back to Step 3.2 (next WHILE iteration)
|
|
225
|
+
ELSE:
|
|
226
|
+
Display: " Max attempts reached. Escalating."
|
|
227
|
+
GOTO Step 7 (Escalation)
|
|
228
|
+
|
|
229
|
+
// Log attempt
|
|
230
|
+
autonomous_log.append({
|
|
231
|
+
phase: phase_number,
|
|
232
|
+
attempt: attempt,
|
|
233
|
+
verifier: verifier_verdict,
|
|
234
|
+
reviewer: reviewer_verdict,
|
|
235
|
+
warrior_score: score,
|
|
236
|
+
result: phase_result
|
|
237
|
+
})
|
|
238
|
+
|
|
239
|
+
// ──────────────────────────────────────────────────────
|
|
240
|
+
// Step 3.5: Phase Transition
|
|
241
|
+
// ──────────────────────────────────────────────────────
|
|
242
|
+
|
|
243
|
+
IF phase_complete:
|
|
244
|
+
phases_completed += 1
|
|
245
|
+
|
|
246
|
+
Update CONSCIENCE.md:
|
|
247
|
+
Phase {phase_number}: complete
|
|
248
|
+
Autonomous: true
|
|
249
|
+
Attempts: {attempt}
|
|
250
|
+
Verifier: {score}/70
|
|
251
|
+
Reviewer: {reviewer_verdict}
|
|
252
|
+
|
|
253
|
+
// Commit checkpoint
|
|
254
|
+
git add -A && git commit -m "autonomous: Phase {phase_number} - {phase_name} complete"
|
|
255
|
+
|
|
256
|
+
Display: " Phase {phase_number} complete. Advancing..."
|
|
257
|
+
```
|
|
258
|
+
|
|
259
|
+
### Step 4: Blocker Detection
|
|
260
|
+
|
|
261
|
+
```
|
|
262
|
+
STOP AUTONOMOUS MODE immediately if ANY of:
|
|
263
|
+
|
|
264
|
+
1. Circuit breaker TRIPPED with state = BLOCKED
|
|
265
|
+
→ Agent cannot make progress. No amount of retry will help.
|
|
266
|
+
→ Display: "Circuit breaker tripped: {reason}"
|
|
267
|
+
|
|
268
|
+
2. Path verification gate fails
|
|
269
|
+
→ Working directory mismatch or cross-project contamination detected.
|
|
270
|
+
→ Display: "PATH VIOLATION: {details}"
|
|
271
|
+
|
|
272
|
+
3. HAC HARD BLOCK fires (confidence 5/5 rule match)
|
|
273
|
+
→ A known-bad action was attempted.
|
|
274
|
+
→ Display: "HAC BLOCK: Rule {N} — {rule statement}"
|
|
275
|
+
|
|
276
|
+
4. 3 consecutive phase attempts fail (max_phase_attempts exceeded)
|
|
277
|
+
→ Genuine blocker in the phase.
|
|
278
|
+
→ Display: "Phase {N} failed after {max_phase_attempts} attempts"
|
|
279
|
+
|
|
280
|
+
5. Context degradation detected (output volume decline >50%)
|
|
281
|
+
→ Context rot. Fresh session needed.
|
|
282
|
+
→ Display: "Context degradation detected. Sabbath Rest recommended."
|
|
283
|
+
|
|
284
|
+
On ANY blocker: GOTO Step 7 (Escalation)
|
|
285
|
+
```
|
|
286
|
+
|
|
287
|
+
### Step 5: Write Autonomous Log
|
|
288
|
+
|
|
289
|
+
```
|
|
290
|
+
Write .planning/autonomous-log.md:
|
|
291
|
+
|
|
292
|
+
# Autonomous Session Log
|
|
293
|
+
|
|
294
|
+
## Session: {timestamp}
|
|
295
|
+
|
|
296
|
+
| Key | Value |
|
|
297
|
+
|-----|-------|
|
|
298
|
+
| Phases | {from_phase} to {to_phase} |
|
|
299
|
+
| Completed | {phases_completed} of {total} |
|
|
300
|
+
| Duration | {elapsed_time} |
|
|
301
|
+
| Mode | Full autopilot |
|
|
302
|
+
|
|
303
|
+
### Phase Results
|
|
304
|
+
|
|
305
|
+
| Phase | Name | Attempts | Verifier | Reviewer | Result |
|
|
306
|
+
|-------|------|----------|----------|----------|--------|
|
|
307
|
+
| {N} | {name} | {attempts} | {score}/70 | {verdict} | {result} |
|
|
308
|
+
| {N+1} | {name} | {attempts} | {score}/70 | {verdict} | {result} |
|
|
309
|
+
|
|
310
|
+
### Gaps Noted
|
|
311
|
+
{List of non-critical gaps from PASSED WITH NOTES phases}
|
|
312
|
+
|
|
313
|
+
### Behavioral Directive Proposals
|
|
314
|
+
{Any HIGH/CRITICAL findings from reviewer that should become rules}
|
|
315
|
+
```
|
|
316
|
+
|
|
317
|
+
### Step 6: Completion Banner
|
|
318
|
+
|
|
319
|
+
```
|
|
320
|
+
+--------------------------------------------------------------+
|
|
321
|
+
| AUTONOMOUS MODE COMPLETE |
|
|
322
|
+
+--------------------------------------------------------------+
|
|
323
|
+
| |
|
|
324
|
+
| Phases completed: {phases_completed} of {total} |
|
|
325
|
+
| Total attempts: {sum of all attempts} |
|
|
326
|
+
| Duration: {elapsed_time} |
|
|
327
|
+
| |
|
|
328
|
+
| Results: |
|
|
329
|
+
| Phase {N}: {result} (verifier {score}/70, reviewer {verdict}) |
|
|
330
|
+
| Phase {N+1}: {result} (...) |
|
|
331
|
+
| ... |
|
|
332
|
+
| |
|
|
333
|
+
| Gaps noted: {count} |
|
|
334
|
+
| Log: .planning/autonomous-log.md |
|
|
335
|
+
| |
|
|
336
|
+
+--------------------------------------------------------------+
|
|
337
|
+
| NEXT |
|
|
338
|
+
+--------------------------------------------------------------+
|
|
339
|
+
| |
|
|
340
|
+
| Your work is ready for review. |
|
|
341
|
+
| |
|
|
342
|
+
| /fire-dashboard — See project status |
|
|
343
|
+
| /fire-4-verify {N} — Detailed verification |
|
|
344
|
+
| /fire-session-summary — Auto-generated (see above) |
|
|
345
|
+
| /fire-5-handoff — Full WARRIOR handoff (milestones) |
|
|
346
|
+
| |
|
|
347
|
+
+--------------------------------------------------------------+
|
|
348
|
+
```
|
|
349
|
+
|
|
350
|
+
### Step 7: Escalation (Blocker Hit)
|
|
351
|
+
|
|
352
|
+
```
|
|
353
|
+
+--------------------------------------------------------------+
|
|
354
|
+
| AUTONOMOUS MODE STOPPED — BLOCKER |
|
|
355
|
+
+--------------------------------------------------------------+
|
|
356
|
+
| |
|
|
357
|
+
| Completed: Phases {from_phase} to {last_completed} |
|
|
358
|
+
| Blocked at: Phase {blocked_phase}, attempt {attempt} |
|
|
359
|
+
| |
|
|
360
|
+
| Blocker: {description} |
|
|
361
|
+
| Type: {BLOCKED | PATH_VIOLATION | HAC_BLOCK | MAX_ATTEMPTS | |
|
|
362
|
+
| CONTEXT_DEGRADATION} |
|
|
363
|
+
| |
|
|
364
|
+
| What was tried: |
|
|
365
|
+
| - Attempt 1: {verifier_verdict} / {reviewer_verdict} |
|
|
366
|
+
| - Attempt 2: {verifier_verdict} / {reviewer_verdict} |
|
|
367
|
+
| - Attempt 3: {verifier_verdict} / {reviewer_verdict} |
|
|
368
|
+
| |
|
|
369
|
+
| Log: .planning/autonomous-log.md |
|
|
370
|
+
| |
|
|
371
|
+
+--------------------------------------------------------------+
|
|
372
|
+
| OPTIONS |
|
|
373
|
+
+--------------------------------------------------------------+
|
|
374
|
+
| |
|
|
375
|
+
| A) /fire-debug — Investigate the blocker |
|
|
376
|
+
| B) /fire-3-execute {N} — Manual execution with oversight |
|
|
377
|
+
| C) /fire-autonomous --from-phase {N} — Retry from here |
|
|
378
|
+
| D) /fire-dashboard — Review current state |
|
|
379
|
+
| |
|
|
380
|
+
+--------------------------------------------------------------+
|
|
381
|
+
```
|
|
382
|
+
|
|
383
|
+
### Step 7.5: Auto Session Summary
|
|
384
|
+
|
|
385
|
+
```
|
|
386
|
+
Run /fire-session-summary
|
|
387
|
+
|
|
388
|
+
This generates a compact forward-looking summary with:
|
|
389
|
+
- Aggregate status (DONE/PARTIAL/BLOCKED for each work item)
|
|
390
|
+
- Readiness assessment (what's ready, blocked, needs-first)
|
|
391
|
+
- Outlook (trajectory, risk, momentum)
|
|
392
|
+
- Next steps (3 specific actions)
|
|
393
|
+
- Decisions made
|
|
394
|
+
|
|
395
|
+
Saved to ~/.claude/session-summaries/{project}_{date}.md
|
|
396
|
+
Auto-indexed into Qdrant on next session-end consolidation.
|
|
397
|
+
|
|
398
|
+
This is NOT a replacement for WARRIOR handoffs — it's the lightweight
|
|
399
|
+
forward-looking context that the memory system doesn't capture.
|
|
400
|
+
```
|
|
401
|
+
|
|
402
|
+
### Step 8: Sabbath Rest — Session State
|
|
403
|
+
|
|
404
|
+
```
|
|
405
|
+
Update .claude/dominion-flow.local.md:
|
|
406
|
+
|
|
407
|
+
---
|
|
408
|
+
last_session: {timestamp}
|
|
409
|
+
command: "autonomous"
|
|
410
|
+
status: {complete | blocked}
|
|
411
|
+
phases_completed: {list}
|
|
412
|
+
blocked_at: {phase or null}
|
|
413
|
+
blocker_type: {type or null}
|
|
414
|
+
log: ".planning/autonomous-log.md"
|
|
415
|
+
---
|
|
416
|
+
|
|
417
|
+
# Autonomous Session State
|
|
418
|
+
|
|
419
|
+
## Last Run
|
|
420
|
+
- Date: {timestamp}
|
|
421
|
+
- Phases: {from} to {to}
|
|
422
|
+
- Completed: {count}/{total}
|
|
423
|
+
- Duration: {elapsed}
|
|
424
|
+
|
|
425
|
+
## Resume Point
|
|
426
|
+
IF blocked:
|
|
427
|
+
Resume with: /fire-autonomous --from-phase {blocked_phase}
|
|
428
|
+
ELSE:
|
|
429
|
+
All phases complete. Next milestone or new discussions.
|
|
430
|
+
```
|
|
431
|
+
|
|
432
|
+
IF phases_completed > 0:
|
|
433
|
+
Create WARRIOR handoff with autonomous session context.
|
|
434
|
+
|
|
435
|
+
---
|
|
436
|
+
|
|
437
|
+
## Safety Guarantees
|
|
438
|
+
|
|
439
|
+
```
|
|
440
|
+
NEVER DISABLED in autonomous mode:
|
|
441
|
+
|
|
442
|
+
1. Path Verification Gate (Step 0.5)
|
|
443
|
+
— Wrong-repo editing is catastrophic and irreversible
|
|
444
|
+
— No confidence level justifies skipping this
|
|
445
|
+
|
|
446
|
+
2. HAC Hard Blocks (confidence 5/5 anti-patterns)
|
|
447
|
+
— Known-bad actions are blocked BEFORE execution
|
|
448
|
+
— These represent "we know this is wrong" conclusions
|
|
449
|
+
|
|
450
|
+
3. Circuit Breaker (BLOCKED state)
|
|
451
|
+
— Agent admits it cannot proceed
|
|
452
|
+
— No amount of retry will fix external blockers
|
|
453
|
+
|
|
454
|
+
4. Power-Verifier (70-point WARRIOR validation)
|
|
455
|
+
— Functional correctness checking stays active
|
|
456
|
+
— Verdicts auto-route but are never suppressed
|
|
457
|
+
|
|
458
|
+
5. Power-Reviewer (15-persona code review)
|
|
459
|
+
— Quality and simplicity checking stays active
|
|
460
|
+
— BLOCK verdicts force fix cycles, not human pauses
|
|
461
|
+
|
|
462
|
+
WHAT IS DISABLED in autonomous mode:
|
|
463
|
+
|
|
464
|
+
1. "READY FOR HUMAN" pause at merge gate
|
|
465
|
+
— Auto-proceeds to next phase instead of waiting
|
|
466
|
+
|
|
467
|
+
2. "Continue?" interrupts between breaths
|
|
468
|
+
— Handled by --auto-continue flag
|
|
469
|
+
|
|
470
|
+
3. Plan-checker blocking
|
|
471
|
+
— Handled by --skip-checker flag
|
|
472
|
+
— Plan issues surface in verification instead
|
|
473
|
+
|
|
474
|
+
4. Human review of individual phase results
|
|
475
|
+
— Results logged, human sees final product
|
|
476
|
+
```
|
|
477
|
+
|
|
478
|
+
---
|
|
479
|
+
|
|
480
|
+
## Examples
|
|
481
|
+
|
|
482
|
+
```bash
|
|
483
|
+
# Full autopilot from next incomplete phase to end
|
|
484
|
+
/fire-autonomous
|
|
485
|
+
|
|
486
|
+
# Run phases 3 through 7
|
|
487
|
+
/fire-autonomous --from-phase 3 --to-phase 7
|
|
488
|
+
|
|
489
|
+
# Preview without executing
|
|
490
|
+
/fire-autonomous --dry-run
|
|
491
|
+
|
|
492
|
+
# Custom iteration limits
|
|
493
|
+
/fire-autonomous --max-phase-attempts 5 --max-iterations 100
|
|
494
|
+
```
|
|
495
|
+
|
|
496
|
+
---
|
|
497
|
+
|
|
498
|
+
## Related Commands
|
|
499
|
+
|
|
500
|
+
- `/fire-1a-discuss` — Gather context BEFORE autonomous mode (required)
|
|
501
|
+
- `/fire-2-plan` — Manual planning (autonomous mode auto-plans)
|
|
502
|
+
- `/fire-3-execute` — Manual execution (autonomous mode auto-executes)
|
|
503
|
+
- `/fire-4-verify` — Manual verification (autonomous mode auto-verifies)
|
|
504
|
+
- `/fire-loop` — Task-level loop (autonomous mode is phase-level)
|
|
505
|
+
- `/fire-loop-stop` — Emergency stop for autonomous mode
|
|
506
|
+
- `/fire-dashboard` — View results after autonomous run
|
|
507
|
+
|
|
508
|
+
---
|
|
509
|
+
|
|
510
|
+
## Success Criteria
|
|
511
|
+
|
|
512
|
+
- [ ] All target phases have MEMORY.md (prerequisite enforced)
|
|
513
|
+
- [ ] Path verification gate runs on entry
|
|
514
|
+
- [ ] Each phase is auto-planned if no BLUEPRINT.md exists
|
|
515
|
+
- [ ] Each phase is auto-executed with --auto-continue --autonomous
|
|
516
|
+
- [ ] Verification results read and evaluated automatically
|
|
517
|
+
- [ ] FAILED/BLOCK phases auto-replan and re-execute (up to max attempts)
|
|
518
|
+
- [ ] PASSED phases auto-advance to next phase
|
|
519
|
+
- [ ] Autonomous log written with per-phase details
|
|
520
|
+
- [ ] Blocker detection stops autonomous mode cleanly
|
|
521
|
+
- [ ] Escalation banner shows options for manual intervention
|
|
522
|
+
- [ ] WARRIOR handoff created on completion or blocker
|
|
523
|
+
- [ ] Sabbath Rest state updated for resume capability
|
|
524
|
+
|
|
525
|
+
---
|
|
526
|
+
|
|
527
|
+
*Dominion Flow v9.0 — Full autopilot with instrumentation over restriction*
|
|
528
|
+
*Boris Cherny principle: "You don't trust; you instrument."*
|