ワンクリックで
bmad-autopilot-skill
Autonomous BMAD development orchestration
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
メニュー
Autonomous BMAD development orchestration
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Setup Nuxt/Vue/TypeScript project. Use when user needs to install dependencies, prepare Nuxt, setup LSP, or fix TypeScript errors in a Nuxt/Vue project. Works in both main repo and worktrees.
Manage git worktrees. Use when user wants to create, list, delete, or switch worktrees. Helps with branch-based development workflows where each branch has its own working directory.
Maintain the Claude Code plugin marketplace - create plugins, add skills, bump versions, update registry. Use when creating new plugins, adding skills/agents/commands to plugins, updating versions, or managing the marketplace registry.
SSH operations for remote server management, tunneling, and proxy-based connections. Use when connecting to remote servers, running remote commands, setting up SSH tunnels, port forwarding, managing jump hosts/bastions, or executing kubectl/k9s on remote clusters. Triggers on SSH, remote, tunnel, bastion, jump host, port forward, or remote kubernetes access.
Kubernetes troubleshooting and debugging workflows. Use when diagnosing pod failures, networking issues, resource problems, CrashLoopBackOff, ImagePullBackOff, pending pods, OOMKilled, or any cluster issues. Covers systematic debugging approaches and common failure patterns.
Generate Kubernetes manifest templates for deployments, services, configmaps, secrets, ingress, and other resources. Use when creating new k8s resources, scaffolding applications, or needing YAML templates. Triggers on requests to create, generate, or scaffold kubernetes resources.
| name | bmad-autopilot-skill |
| description | Autonomous BMAD development orchestration |
| allowed-tools | Bash,Read |
| version | 1.2.1 |
This skill provides autonomous development capabilities following the BMAD (Breakthrough Method for Agile Development) methodology.
Activate this skill when the user wants to:
CHECK_PENDING_PR → FIND_EPIC → CREATE_BRANCH → DEVELOP_STORIES
↓ ↓
(resume) CODE_REVIEW
↓
DONE ← MERGE_PR ← WAIT_COPILOT ← CREATE_PR ← (approved)
↓ ↓
BLOCKED FIX_ISSUES
# Initialize claude-threads
ct init
# Create BMAD workflow thread
ct thread create bmad-main \
--mode automatic \
--template bmad-autopilot.yaml \
--context '{"epic_pattern": "", "base_branch": "main"}'
# Start the thread
ct thread start bmad-main
# Single epic with isolated worktree
ct thread create bmad-7a \
--template bmad-developer.md \
--worktree \
--context '{"epic_id": "7A"}'
# Multiple epics in parallel (each gets own worktree)
ct thread create bmad-7a --template bmad-developer.md --worktree --context '{"epic_id": "7A"}'
ct thread create bmad-8a --template bmad-developer.md --worktree --context '{"epic_id": "8A"}'
ct thread create bmad-10b --template bmad-developer.md --worktree --context '{"epic_id": "10B"}'
# Pattern matching
ct thread create bmad-all-10 \
--template bmad-autopilot.yaml \
--context '{"epic_pattern": "10.*"}'
# Check thread status
ct thread status bmad-main
# View logs
ct thread logs bmad-main -f
# List all BMAD events
ct event list --type "EPIC_*,STORY_*,PR_*"
# List active worktrees
ct worktree list
The BMAD workflow spawns specialized agent threads:
| Agent | Template | Purpose |
|---|---|---|
| Coordinator | planner.md | Find epics, manage workflow |
| Developer | bmad-developer.md | Implement stories with TDD |
| Reviewer | bmad-reviewer.md | Code review before PR |
| PR Manager | bmad-pr-manager.md | Create/manage pull requests |
| Fixer | bmad-fixer.md | Fix CI/review issues |
| Monitor | pr-monitor.md | Watch for PR status changes |
The BMAD workflow includes two review stages:
After all stories in an epic are implemented, the Reviewer agent performs an internal code review before creating a PR.
Review Checklist:
Review Outcomes:
The reviewer publishes a REVIEW_COMPLETED event with status and any issues found.
After PR creation, the workflow waits for:
Auto-Approval Conditions:
PRs are auto-approved when:
Review Commands:
# View all changes
gh pr diff <pr_number>
# View specific file
gh pr diff <pr_number> -- path/to/file
# Check CI status
gh pr checks <pr_number>
# View PR details
gh pr view <pr_number>
Events published during BMAD execution:
| Event | Description |
|---|---|
NEXT_EPIC_READY | Coordinator found next epic |
BRANCH_CREATED | Feature branch created |
WORKTREE_CREATED | Isolated worktree created |
STORY_STARTED | Developer starting story |
STORY_COMPLETED | Story implementation done |
DEVELOPMENT_COMPLETED | All stories finished |
REVIEW_COMPLETED | Code review done |
WORKTREE_PUSHED | Changes pushed from worktree |
PR_CREATED | Pull request opened |
CI_PASSED / CI_FAILED | CI status update |
COPILOT_REVIEW | Copilot review received |
FIXES_NEEDED | Issues require fixing |
PR_APPROVED | PR approved |
PR_MERGED | PR merged successfully |
WORKTREE_DELETED | Worktree cleaned up |
EPIC_COMPLETED | Epic fully processed |
In .claude-threads/config.yaml:
bmad:
epic_pattern: "" # Process all, or specify "7A 8A"
base_branch: main
auto_merge: true # Merge when approved
max_concurrent_prs: 2 # Limit parallel PRs
check_interval: 300 # PR check interval (seconds)
worktrees:
enabled: true # Enable worktree isolation
max_age_days: 7 # Auto-cleanup after N days
auto_cleanup: true # Cleanup when thread completes
default_base_branch: main # Default base for new worktrees
auto_push: true # Push changes on completion
pr_shepherd:
max_fix_attempts: 5 # Max fix attempts per PR
ci_poll_interval: 30 # CI check interval
auto_merge: false # Auto-merge when approved
BMAD stories are typically in:
_bmad-output/stories/epic-{id}/docs/stories/{id}/feat(epic-{id}): implement story X.Y
Story {id}.{story}:
- Change description
- Another change
If autopilot becomes blocked:
# Check why
ct thread status bmad-main --verbose
# View recent events
ct event list --source bmad-main --limit 20
# Check logs
ct thread logs bmad-main
# Resume manually
ct thread resume bmad-main
For automatic PR status updates:
# Start webhook server
ct webhook start --port 31338
# GitHub webhook settings:
# URL: http://your-server:31338/webhook
# Events: Pull requests, Check runs, Reviews
The project has an auto-approve.yml workflow that handles PR approval automatically.
# Trigger auto-approve for a specific PR
gh workflow run auto-approve.yml -f pr_number=<PR_NUMBER>
# Or use the manual approve-pr workflow with options
gh workflow run approve-pr.yml -f pr_number=<PR_NUMBER> [-f skip_ci_check=false] [-f comment="Optional message"]
PRs are auto-approved when:
IMPORTANT: Do NOT wait passively for review comments. Proactively check:
# Check for unresolved review threads IMMEDIATELY after Copilot review
gh api graphql -f query='query {
repository(owner: "hanibalsk", name: "property-management") {
pullRequest(number: <PR_NUMBER>) {
reviewThreads(first: 100) {
nodes { isResolved path comments(first: 1) { nodes { body } } }
}
}
}
}' | jq '[.data.repository.pullRequest.reviewThreads.nodes[] | select(.isResolved == false)]'
# If unresolved > 0, spawn review-comment-handler agent immediately
# DO NOT wait for the auto-approve workflow to fail
# Combined status check for a PR
gh pr view <PR_NUMBER> --json mergeable,mergeStateStatus,reviewDecision,statusCheckRollup | jq '{
mergeable,
mergeStateStatus,
reviewDecision,
ci_in_progress: [.statusCheckRollup[] | select(.status == "IN_PROGRESS")] | length,
ci_failures: [.statusCheckRollup[] | select(.conclusion == "FAILURE")] | length
}'
Sequential (default):
Parallel (with worktrees):
max_concurrent_prs# Parallel development with worktrees
ct thread create epic-7a --mode automatic --template bmad-developer.md --worktree --context '{"epic_id": "7A"}'
ct thread create epic-8a --mode automatic --template bmad-developer.md --worktree --context '{"epic_id": "8A"}'
ct orchestrator start
# Each thread works in isolation
ct worktree list
The PR Shepherd automatically handles CI failures and review comments with worktree isolation:
# Watch a PR - creates isolated worktree for fixes
ct pr watch 123
# Shepherd automatically:
# - Creates worktree for PR branch
# - Spawns fix threads when CI fails
# - Fix threads work in isolated worktree
# - Pushes fixes from worktree
# - Cleans up worktree when PR merges
ct pr status 123
ct pr daemon # Run as background service