メニュー
Project status briefing for Look Ma No Hands. Reads git activity, open PRs and issues, build health, and last-session handoff to produce a unified briefing. Only invoke when the user explicitly requests a status briefing.
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
Single-ticket implementation chain for Look Ma No Hands GitHub issues. Fetches the issue, assesses scope, routes to fast-path or standard-path, dispatches to engineering agents, and drives the review-fix-merge loop.
End-of-session summary for Look Ma No Hands. Reviews the conversation, writes session log, updates handoff file, and commits wrap-up files. Use at the end of every session.
Manage structured YAML expertise files as personal mental models. Use when starting tasks (read for context), completing work (capture learnings), or when your understanding of the system needs updating.
Quick reference for git operations, commit patterns, and workflow best practices
Analyze session transcript for corrections, approvals, and observations; append new learnings to project memory
Analyzes and optimizes Claude Code project configurations (CLAUDE.md, agents, skills directories) based on Vercel's agent eval research. Use when you want to improve agent performance in a repo by applying passive-context patterns, generating compressed documentation indexes, or auditing existing skill/agent setups for efficiency. Triggers on requests to "optimize my Claude setup", "improve agent performance", "audit my skills directory", or "apply AGENTS.md patterns".
| name | lmnh-heartbeat |
| description | Project status briefing for Look Ma No Hands. Reads git activity, open PRs and issues, build health, and last-session handoff to produce a unified briefing. Only invoke when the user explicitly requests a status briefing. |
Invoke only on explicit user request. Do not run this automatically at session start.
Project status briefing. Executes in two phases: data collection (Scout) and synthesis (Orchestrator).
Spawn a Scout agent with this task:
Collect LMNH project status data. Return results in a ---HEARTBEAT-DATA--- block with these sections:
## last-session
Read memory/last-session.md -- full contents (or "empty" if file is blank).
## recent-commits
git -C /Users/qaid/Code/look-ma-no-hands log --oneline -15
## open-prs
gh pr list --repo qaid/look-ma-no-hands --state open --json number,title,headRefName,updatedAt
## open-issues
gh issue list --repo qaid/look-ma-no-hands --state open --json number,title,labels,updatedAt --limit 30
## build-health
swift build -c release 2>&1 | tail -15
## test-health
swift test 2>&1 | tail -20
## worktrees
git worktree list
---HEARTBEAT-DATA---
[structured output here]
---END HEARTBEAT-DATA---
Use the returned block as-is for Phase 2. Do NOT re-fetch any sources.
Parse the ---HEARTBEAT-DATA--- block returned by Scout and produce the unified briefing below.
[Only if last-session has content]
swift build -c release status: clean / warnings / errorsswift test status: passing / failing / countupdatedAt date for each so staleness is visible