agentleFS
HarambeeAIHarambeeAI/bayes-ship/skills/ship-auto/SKILL.md

ship-auto

Skill1 starsChanged 7 months agoMITView on GitHub

Autonomous build mode. After spec and architecture are approved, runs the entire build pipeline (worktree, plan, execute, test, finish-branch) without stopping. Logs all autonomous decisions. Use --checkpoints for semi-autonomous mode.

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 4 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Create a git worktree

    line 22
  2. Distinctive:

    Run TDD + two-stage review on every task (non-negotiable)

    line 24
  3. Distinctive:

    Run full QA verification

    line 25
  4. Very specific:

    All decisions logged to .claude/state/auto-decisions.md I will only stop if genuinely BLOCKED.

    line 29
  5. Distinctive:

    Every auto-decision is logged with context and confidence

    line 89

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Auto — Autonomous build mode

Usage

/bayes-ship:ship-auto — fully autonomous /bayes-ship:ship-auto --checkpoints — semi-autonomous (pauses between phases)

Pre-condition

Both .claude/state/PRD.md and .claude/state/ARCHITECTURE.md must exist and be approved (check progress.json). Autonomous mode only covers the BUILD phase. Discovery, research, spec, and architecture always require human input.

Process

Announce to the user:

"Autonomous build activated. I will:

  1. Create a git worktree
  2. Plan and execute each phase
  3. Run TDD + two-stage review on every task (non-negotiable)
  4. Run full QA verification
  5. Merge to main
  6. Deploy (if credentials available)

All decisions logged to .claude/state/auto-decisions.md I will only stop if genuinely BLOCKED. Starting now..."

Orchestration loop

1. Create worktree (same as /bayes-ship:ship-worktree)

for each phase in ROADMAP:
    2. Plan the phase (same as /bayes-ship:ship-plan)
       → No user approval needed. Spec + architecture are the contract.

    3. Execute all tasks (same as /bayes-ship:ship-execute)
       → TDD mandatory. Two-stage review mandatory.
       → Atomic commits per task.

    4. Handle NEEDS_CONTEXT autonomously:
       → Check ARCHITECTURE.md and RESEARCH.md for the answer
       → If found: use it, log to auto-decisions.md
       → If not: pick most conservative/standard option, log it
       → Format: { context, sources_checked, decision, confidence, phase, task }

    5. Handle BLOCKED:
       → Attempt workaround (retry with different approach, max 2)
       → If still blocked: STOP and alert user. Save progress.

    6. Verify phase (all tests pass, typecheck clean)
       → If fails: re-plan failed tasks, re-execute (max 2 retries)
       → If still failing: STOP and alert user.

    [Semi-autonomous checkpoint — only with --checkpoints]:
       "Phase N complete: X tasks, Y commits, Z auto-decisions.
        See auto-decisions.md. Continue? (yes / stop)"

after all phases:
    7. Run /bayes-ship:ship-test (full QA)
    8. Run /bayes-ship:ship-finish-branch → auto-merge (squash)
    9. If deploy credentials in environment → /bayes-ship:ship-deploy
       If not → "Build complete. Run /bayes-ship:ship-deploy with credentials."

auto-decisions.md format

# Auto-Decisions Log

## Decision N: [Title]
- Context: [What the implementer asked]
- Sources checked: [Which files were consulted]
- Decision: [What was decided]
- Confidence: High/Medium/Low
- Phase: N, Task: M

Safety guarantees

  • TDD is NEVER skipped in autonomous mode
  • Two-stage review is NEVER skipped in autonomous mode
  • BLOCKED always stops and alerts the user
  • Every auto-decision is logged with context and confidence
  • Verification-before-completion runs at the end

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.