🔄ralph-claude-code
Autonomous AI development loop for Claude Code with intelligent exit detection
⭐ Popular: 10k stars, gaining about 17 a week
View on GitHub ↗repo profile
momentum
durability
bus factor = how many people it takes to cover more than half the commits (6 months). 1 is a solo project; higher means the work is spread across a team. top-author share is the single busiest author's slice of those commits.
since we covered it
why it's a big deal
- Wraps Claude Code in an autonomous loop so it iterates on a project unattended, aimed at developers who want continuous AI-driven work without babysitting each run.
- The main problem it targets is runaway autonomy: it guards against infinite loops and API overuse with rate limiting of 100 calls per hour and a circuit breaker that halts stuck runs.
- It knows when to stop via a dual-condition exit check, requiring both heuristic completion signals and Claude's explicit EXIT_SIGNAL before ending, which avoids premature exits and wasted iterations.
under the hood
- Written almost entirely in Bash and driven by a five-phase loop that reads instructions from.ralph/PROMPT.md, runs Claude Code, tracks progress, evaluates completion, then repeats or exits.
- Depends on the Claude Code CLI plus tmux, jq, Git, and GNU coreutils, and keeps per-project config, logs, and state in a.ralph/ subfolder.
- Handles the 5-hour API limit with three-layer detection (timeout guard, JSON parsing, text fallback), preserves session context across iterations with a 24-hour expiration, and adds GitHub issue integration and batch queue processing.
Radar summary, generated from the project's public sources
star history
- PR#25 3k 2026-01-14
- now 10k + 7k since first covered
curve is sampled from GitHub's star history, plus our own daily readings since we covered it; the dashed stretch is before we first covered it, the solid line since. figures at coverage are the numbers we printed then (approx.), current count is live.
understory
Better known than its recent output, coasting a little on attention.
- output, commits & releases
- clout, star velocity
output = commits & releases; clout = star velocity, both 0 to 100 monthly indices; the gap where output runs above clout is the understory. The understory →
covered in
-
Autonomous dev loop for Claude Code
similar projects
compare these →- ☤ hermes-agent
Python · 24× the stars
The agent that grows with you
231k ACTIVE - 🛠️ Claude Code PM
Project management skill system for Agents that uses GitHub Issues and Git worktrees for parallel agent execution.
8k ACTIVE - 💾 claude-mem
JavaScript · 9.5× the stars
Persistent Context Across Sessions for Every Agent, Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More
91k ACTIVE
comments
Sign in with GitHub to add your blip on ralph-claude-code.