Agent Skills: Ralph Wiggum AI

Ralph Wiggum AI

Autonomous agent that implements features one at a time from a PRD.

Required Files

• PRD: ./plans/prd.yaml
• PROGRESS: ./plans/progress.md
• RESEARCH: ./plans/research.md

ONLY WORK ON A SINGLE FEATURE FROM THE PRD.

Workflow

1. Review RESEARCH for relevant context.
2. Find work: if any task has status: in_progress, continue it. Otherwise, pick the highest-priority pending task — prioritize by YOUR judgment, not list order.
3. Set the task's status to in_progress in the PRD.
4. Implement the feature.
5. Lint and test: if failures occur, attempt to fix and retry ONCE. If still failing, set status to blocked, document the blocker in PROGRESS, and stop.
6. UI verification (auto-triggered for category: ui tasks):
   • Start dev server using dev_command from PRD if base_url is unreachable
   • Open the page: agent-browser open <base_url><path> (add --headed if task has headed: true)
   • For each verification step:
     • agent-browser snapshot — assert expected elements/state from the accessibility tree
     • agent-browser screenshot — save to plans/screenshots/<task-slug>.png as visual evidence
   • Close when done: agent-browser close
7. Set the task's status to done in the PRD.
8. Append to PROGRESS — leave a useful note for the next session.
9. Commit only the changed files with a concise conventional commit message.
10. If all PRD tasks are done, output <promise>COMPLETE</promise> and stop.