Auto Mode

Fully autonomous development. Works through all tasks without stopping until complete.

Entry Flow

auto
  ├─ Check prd.json exists?
  │   ├─ No → Bootstrap from context
  │   └─ Yes → Check pending tasks
  │             ├─ None pending → All done!
  │             └─ Has pending → Execute tasks
  │
  └─ Execute until done or interrupted

CRITICAL: NEVER STOP

FORBIDDEN:

• •NEVER use AskUserQuestion - make decisions yourself
• •NEVER ask "Should I continue?"
• •NEVER show summaries and wait
• •NEVER say "Let me know..."
• •NEVER output minimal responses (., —, Idle.)

REQUIRED:

• •Make autonomous decisions
• •Log decisions to .claude/decisions.md
• •Keep working until truly done

Bootstrap (No prd.json)

When prd.json doesn't exist:

1. •Read CLAUDE.md, README.md, package.json for context
2. •Generate 5-10 starter tasks based on project
3. •Create prd.json with stories
4. •Continue immediately - don't stop for approval

Pre-flight (Quick)

Before first task:

git status --short          # Warn if dirty, continue anyway
npm run build 2>&1 | tail -5  # Fail if broken, fix first

Skip if takes >10 seconds.

Task Execution

Find Next Task

// Find executable tasks (not done, not blocked)
const executable = stories.filter(s =>
  s.passes !== true &&
  (s.blockedBy || []).every(dep =>
    stories.find(d => d.id === dep)?.passes === true
  )
);

Execute Each Task

1. •Read the task description
2. •Implement the solution
3. •npm run typecheck - Fix if fails
4. •npm run build - Fix if fails
5. •Verify (see below)
6. •Update prd.json: passes: true
7. •IMMEDIATELY start next task

Verification

For UX tasks - browser check required:

agent-browser open http://localhost:3000/path
agent-browser snapshot -i  # Verify expected element

Parallel Execution (Optional)

For independent tasks, launch multiple agents:

Task({ subagent_type: "general-purpose", prompt: "...", run_in_background: true })
Task({ subagent_type: "general-purpose", prompt: "...", run_in_background: true })

Smart Retry

On failure:

1. •Log to .claude/mistakes.md
2. •Retry 1: Different approach
3. •Retry 2: Simplest implementation
4. •Still fails → passes: false, continue to next

Commit Cadence

• •Commit every 3 completed tasks
• •Or after major milestones
• •Use conventional commits: feat|fix|refactor

Auto-Checkpoint (Token Protection)

After every 3 completed tasks, save checkpoint and recommend /compact:

if (completedThisSession % 3 === 0) {
  Write checkpoint to .claude/checkpoint.md

  Output:
  "💾 Checkpoint saved. Run /compact to reclaim ~40% tokens.
   Use /clear only at major transitions (~70% but wipes context)."
}

Be concise. Long responses burn tokens. Short responses = more runway.

Completion

When stories.every(s => s.passes === true):

All [N] tasks complete.

Summary:
- [X] features implemented
- [X] bugs fixed
- [X] improvements made

Run `status` to see full results.

IDLE Detection

If no tasks to work on:

1. •Check: Are ALL stories passes: true?
   • •YES → Output completion summary
   • •NO → Find blocked tasks and resolve blockers

Quick Reference