Survey

Runs a structured research pass before you commit to building something. It freezes one question, searches four parallel lanes (context, solutions, actual behavior, alternatives or platform differences), and writes reusable markdown artifacts into .survey/{slug}/ with a fixed contract you can validate. Best when you need to understand what exists, how people work around it, or how Claude/Codex/Gemini handle the same capability differently before deciding what to build. Not for implementation work or small fixes. The artifact structure is strict on purpose: triage, context, solutions, and platform-map when relevant. It stays in research mode, keeps claims source-backed, and hands off to planning or execution only after the landscape is clear.

Install to Claude Code

npx -y skills add akillness/oh-my-skills --skill survey --agent claude-code