elves
Autonomous multi-batch development agent for long unattended runs. Takes a plan, breaks it into sprint-sized batches, implements with testing and PR-based review, and documents everything for compaction recovery. Use when user says "run overnight", "I'm going offline", "implement this plan", "keep going without me", "do not stop", "I'll be back in the morning", "run this end-to-end", or any indication of autonomous execution. Also use when bootstrapping a new project for overnight runs — the skill generates survival guides and execution logs from templates.
npx skills add majiayu000/claude-skill-registry --skill elves --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
What it does
Autonomous batch-driven development agent that takes a plan, splits it into sprint-sized batches, implements each batch with testing and PR-based review, and documents everything to support compaction recovery. It runs while the user is offline, executing end-to-end work in multiple batches, and generates a survival guide and an execution log from templates for overnight runs or bootstrapping new projects.
How it works
- Starts in Phase 1: Planning with the user to produce a Plan, Survival guide, and Execution log, decomposed into batches.
- Runs in finite or open-ended modes as configured. Open-ended mode continues until the user stops or a genuine blocker occurs.
- For each batch: create a roll-forward plan, implement code, run tests, process PR-based review, and update documentation.
- After each batch, read PR comments, fix issues, push updates, and iterate until the batch passes validation and review, then proceed to the next batch.
- Maintains memory across compactions using Survival Guide and Execution Log; ensures context is preserved for long unattended runs.
- Never merges; the user merges when they return.
When to use it
Use when the user anticipates being offline or away for an extended period (e.g., overnight), needs to implement a plan autonomously, or is bootstrapping a new project for overnight runs. Trigger phrases include: "run overnight", "I'm going offline", "implement this plan", "keep going without me", "do not stop", "I'll be back in the morning", "run this end-to-end".
What it can touch
Requires tools and environment as declared: it uses git and the gh CLI for branch handling, planning, PR creation, and reviews. It interacts with the repository to push commits, create and view PRs, and apply tags for batches (e.g., git tag elves/pre-batch-N). It expects compatibility with Claude Code, Codex, Claude.ai, and any agent skills on the platform.
Caveats
The skill is scaffolding and requires project-specific customization for survival guide, test gates, and review processes. It is designed to operate without merging (the user merges when they return). It relies on persistent templates for survival guide and execution log and assumes the existence of a plan decomposed into batches. Some checks and configurations are environment-dependent (GitHub auth, remote availability, and tool settings).
# Elves You are the night shift. The user is the day manager handing you written notes before going offline. Your job is to execute plan-driven work autonomously, batch by batch, with testing, review, and documentation, until the plan is complete or you hit a genuine blocker. **You never merge. The user merges when they return.** **This skill is scaffolding.** It gives you a framework: the loop, the documents, the gates. But every project is different. The user will customize the survival guide, the test gates, and the review process for their specific needs. Follow the framework, but adapt to what the project actually requires. ## Why This Exists Your user has 12 to 14 hours each day when they aren't working: evenings, nights, weekends. You are the mechanism that converts those idle hours into shipped code. The user plans during the day and hands you written notes before going offline. You execute while they sleep. When they return, finished work is waiting. Your core pattern is the Ralph Loop: try, check, feed back, repeat. You don't return correct or incorrect answers. You return drafts. Each batch is a draft that gets refined through validation and review until it passes. A dum
- Why This Exists
- Run Mode
- Open-ended rules
- Pre-Final Guard
- Phase 1: Planning (Interactive)
- What to talk about
- What this phase produces
- Preflight
- Time Awareness
- Setup: Branch, Plan, PR
- Batch Decomposition
- Subagent Strategy
- Core Loop
- 1. Orient
git checkout -b feat/<name-from-plan> git add <survival-guide> <execution-log> <plan-if-new> git commit -m "docs: elves session setup — survival guide, execution log, batch plan" git push -u origin HEAD gh pr create --title "<concise title from plan>" --body "<plan summary with batch list>" gh pr view --json number -q .number git rm <survival-guide-path> <execution-log-path> .elves-session.json git commit -m "chore: remove elves session artifacts from PR"
What does the elves skill do?
Autonomous multi-batch development agent for long unattended runs. Takes a plan, breaks it into sprint-sized batches, implements with testing and PR-based review, and documents everything for compaction recovery. Use when user says "run overnight", "I'm going offline", "implement this plan", "keep going without me", "do not stop", "I'll be back in the morning", "run this end-to-end", or any indication of autonomous execution. Also use when bootstrapping a new project for overnight runs — the skill generates survival guides and execution logs from templates.
How do I install it?
Run `npx skills add majiayu000/claude-skill-registry --skill elves --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From majiayu000/claude-skill-registry, a repository with 534 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.