Skip to main content
ClaudeWave
Skill904 repo starsupdated yesterday

vc-autopilot

vc-autopilot is a contract skill that manages the provisional goal block artifact for Autopilot Mode sessions in the vibecode-pro-max-kit framework. It owns the 9-field goal block format, handles emission after clarification rounds, detects resume scenarios from pasted blocks, and coordinates with the canonical autopilot protocol. Use this skill when the orchestrator needs to emit a new goal block, validate an existing one against the specification, or determine whether an incoming goal block represents a session resume.

Install in Claude Code
Copy
git clone --depth 1 https://github.com/withkynam/vibecode-pro-max-kit /tmp/vc-autopilot && cp -r /tmp/vc-autopilot/.claude/skills/vc-autopilot ~/.claude/skills/vc-autopilot
Then start a new Claude Code session; the skill loads automatically.

SKILL.md

# vc-autopilot

Contract skill that owns the **provisional goal block artifact** for Autopilot Mode sessions. This skill is invoked by the orchestrator to emit, validate, and resume from the goal block. The canonical protocol lives in `process/development-protocols/autopilot.md`.

---

## When To Invoke

Invoke this skill when:

- The orchestrator has just received an autopilot trigger phrase and needs to emit the provisional goal block after the consolidated clarification round.
- The orchestrator detects a pasted goal block at session start and needs to determine whether it is a resume scenario (goal block already present → no new clarification round).
- The `validate-autopilot-goal-block.mjs` D1 validator needs to be run against an artifact.

---

## Skill Artifact

The **provisional goal block** — a structured text block of ≤ 4000 characters with exactly 9 named fields. The canonical field spec lives in `process/development-protocols/autopilot.md §Provisional Goal Block Format`. This skill does not redefine the spec — it references it.

Required fields (exact string anchors — do not rename or abbreviate):

1. `SESSION GOAL:`
2. `ENTRY PHASE:`
3. `REMAINING PHASES:`
4. `CLARIFICATIONS LOCKED:`
5. `EXECUTE CONSENT:` — must contain the literal text `standing-granted`
6. `DECISION POLICY:`
7. `HARD STOPS:`
8. `TEST GATES:`
9. `START:`

---

## Emission Procedure

Step-by-step for the orchestrator:

1. Complete the consolidated clarification round (`process/development-protocols/autopilot.md §Consolidated Clarification Round`).
2. Determine `ENTRY PHASE` from on-disk artifact detection (autopilot.md §Trigger-Anywhere Detection Flow).
3. Build the `REMAINING PHASES` checklist: for each phase not yet complete in canonical RIPER-5 order, add a `[ ]` checkbox line with the phase name and planned execution strategy.
4. Fill in all 9 fields using the locked clarification answers.
5. Count total characters. If > 4000: compress DECISION POLICY and CLARIFICATIONS LOCKED to summaries and reference autopilot.md for full detail.
6. Print the block to chat as a fenced code block.
7. Write the block to disk: `{task-folder}/{slug}_AUTOPILOT_GOAL_{dd-mm-yy}.md` (header: "Emitted: [datetime]. Provisional block. V7 will emit (UPDATE) variant.").
8. Emit `AUTOPILOT_ACTIVATED: [task] — entry phase: [phase] — goal block emitted`.

---

## Resume Detection

How the orchestrator recognizes a pasted goal block at session start:

- If the user message contains all 9 field names as headings/labels, treat it as an autopilot resume.
- On resume: skip the consolidated clarification round entirely. Read `ENTRY PHASE` and `REMAINING PHASES` from the pasted block. Read `CLARIFICATIONS LOCKED` as the already-locked decisions. Read `DECISION POLICY` and `HARD STOPS` as the standing policy.
- Emit `[MODE: AUTOPILOT | <ENTRY PHASE>]` and begin the run from `START:`.
- Do NOT issue a new clarification round (SPEC AC-14).

---

## V7 UPDATE Variant

When VALIDATE V7 completes during an autopilot run, the orchestrator:

1. Reads the real gate commands from the new validate-contract.
2. Copies the provisional block.
3. Prefixes `SESSION GOAL:` with `(UPDATE) `.
4. Replaces `TEST GATES: TBD — populated after VALIDATE` with the actual gate commands.
5. Updates `START:` to reflect post-VALIDATE state.
6. Prints the updated block to chat.
7. Overwrites the disk artifact at the same path.

---

## Validator

Run after writing any goal block artifact to confirm it passes D1 checks:

```bash
node .claude/skills/vc-autopilot/scripts/validate-autopilot-goal-block.mjs <artifact-path>
```

Exit 0 = PASS (all 9 required fields present, EXECUTE CONSENT contains `standing-granted`, total ≤ 4000 chars).
Exit 1 = FAIL (one or more checks failed — error printed to stdout).
WARN printed (exit 0) when `TEST GATES:` contains `TBD` (reminder that V7 UPDATE is pending).

See `scripts/validate-autopilot-goal-block.mjs` and `fixtures/` for the D1 validator and pass/fail fixture pair.
code-reviewerSubagent

Comprehensive code review with scout-based edge case detection. Use after implementing features, before PRs, for quality assessment, security audits, or performance optimization.

code-simplifierSubagent

Simplifies and refines code for clarity, consistency, and maintainability while preserving all functionality. Focuses on recently modified code unless instructed otherwise.

debuggerSubagent

Use this agent when you need to investigate issues, analyze system behavior, diagnose performance problems, examine database structures, collect and analyze logs from servers or CI/CD pipelines, run tests for debugging purposes, or optimize system performance. This includes troubleshooting errors, identifying bottlenecks, analyzing failed deployments, investigating test failures, and creating diagnostic reports. Examples:\n\n<example>\nContext: The user needs to investigate why an API endpoint is returning 500 errors.\nuser: "The /api/users endpoint is throwing 500 errors"\nassistant: "I''ll use the debugger agent to investigate this issue"\n<commentary>\nSince this involves investigating an issue, use the Task tool to launch the debugger agent.\n</commentary>\n</example>\n\n<example>\nContext: The user wants to analyze why the CI/CD pipeline is failing.\nuser: "The GitHub Actions workflow keeps failing on the test step"\nassistant: "Let me use the debugger agent to analyze the CI/CD pipeline logs and identify the issue"\n<commentary>\nThis requires analyzing CI/CD logs and test failures, so use the debugger agent.\n</commentary>\n</example>\n\n<example>\nContext: The user notices performance degradation in the application.\nuser: "The application response times have increased by 300% since yesterday"\nassistant: "I''ll launch the debugger agent to analyze system behavior and identify performance bottlenecks"\n<commentary>\nPerformance analysis and bottleneck identification requires the debugger agent.\n</commentary>\n</example>

execute-agentSubagent

EXECUTE MODE - Implementing EXACTLY what was planned. Full tool access. Can only be invoked after explicit user confirmation. Use after plan is approved.

fast-mode-agentSubagent

FAST MODE - Execute compressed RIPER-5 workflow (RESEARCH + INNOVATE + PLAN) in one session, then pause for EXECUTE confirmation. Use when you want quick end-to-end solution.

git-managerSubagent

Stage, commit, and push code changes with conventional commits. Use when user says "commit", "push", or finishes a feature/fix.

innovate-agentSubagent

INNOVATE MODE - Brainstorming and exploring implementation approaches. Discusses possibilities without making decisions. Use after research is complete.

plan-agentSubagent

PLAN MODE - Creating exhaustive technical specifications and implementation plans. Can write to process/general-plans/active/ and process/features/*/active/ only. Use after approach is decided.