2.5k
GitHub Stars
1
Bundled Files
2 months ago
Catalog Refreshed
4 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill openclaw/skills --skill ralph-skill- SKILL.md10.5 KB
Overview
This skill automates PRD-driven development using the RALPH BUILD LOOP and Claude Code. It creates structured PRDs, manages prioritized tasks, runs tests, and performs git commits so a project can be developed automatically and audibly tracked. Use it to bootstrap new projects and keep development aligned with acceptance criteria and security-first defaults.
How this skill works
Ralph initializes a PRD.json and project baseline (.gitignore, .env.example, ralph.config.json) then starts Claude Code with the PRD to execute an automated 8-step build loop. It reads project files, implements tasks per acceptance criteria, runs tests, commits changes with task IDs, updates task status in the PRD, and repeats until completion. A background monitor reports session status, file changes, recent activity, and alerts if the session stops unexpectedly.
When to use it
- Starting a new project and you want structured, PRD-driven automation.
- Bootstrapping a repository with security-first defaults (.gitignore, .env.example).
- Delegating iterative implementation, testing, and committing to Claude Code.
- Tracking task progress and dependencies automatically during development.
- Running language-specific test suites after each implemented task.
Best practices
- Edit the generated PRD.json to reflect your exact acceptance criteria before running a build.
- Keep sensitive operations guarded; allow Ralph to manage safe approvals but review prompts when unsure.
- Use --auto-commit for reproducible history and clear task-based commit messages.
- Start with the default security section and confirm .env.example and .gitignore are correct.
- Configure test_command in ralph.config.json for non-standard test setups.
Example use cases
- Initialize a Python API project: ralph init --name "Todo API" --language python then ralph build --auto-commit.
- Monitor a long-running Claude Code session: ralph monitor --session <session_id> --dir <project_dir> --interval 60.
- Create CI/CD and GitHub integration during setup: ralph init --name "Project" --github.
- Resume work from a specific section after fixing tests: ralph build --section 04_test.
- Enforce task dependencies so database setup waits for core infra tasks to complete.
FAQ
Ralph pauses the loop, reports the test output, and updates the PRD status so you can fix failures before resuming the build.
Can I customize the test command?
Yes. Set test_command in ralph.config.json to override defaults like pytest or npm test.