ralph-wiggum
Spec-driven autonomous AI coding with fresh context per task using iterative bash loops.
- Implements Geoffrey Huntley's methodology where each loop iteration starts a new agent process with a clean context window, preventing degradation over long sessions
- Requires clear, testable acceptance criteria in specification files; agent outputs
<promise>DONE</promise>only when all criteria are verified and tests pass - Maintains shared state on disk via
specs/,ralph_history.txt, and optionalIMPLEMENTATION_PLAN.mdfor persistence across iterations - Supports two modes: build (default, for implementation) and plan (for task breakdown); works with Claude Code and Codex CLI
Ralph Wiggum
Autonomous AI coding with spec-driven development
What is Ralph Wiggum?
Ralph Wiggum combines Geoffrey Huntley's iterative bash loop with spec-driven development for fully autonomous AI-assisted software development.
The key insight: Fresh context each iteration. Each loop starts a new agent process with a clean context window, preventing context overflow and degradation.
When to Use This Skill
Use Ralph Wiggum when:
- You have multiple specifications/features to implement
- You want the AI to work autonomously through tasks
- You need consistent, verifiable completion of acceptance criteria
- You want to avoid context window problems in long sessions
More in Agent workflows
find-skills
Discover and install skills from skills.sh directly inside an agent session
vercel-labs/skillsagent-browser
Full browser automation: navigate, click, fill forms, extract data, and screenshot
vercel-labs/agent-browserskill-creator
Create, test, and publish new skills from within your agent
anthropics/skillsbrainstorming
Structured ideation and problem decomposition frameworks
obra/superpowersbrowser-use
Browser automation with visual understanding — interacts with pages based on what it sees
browser-use/browser-use