Skill Detail

Run Genie planning and execution loops in Claude Code

Use Genie to turn a vague coding request into Claude Code brainstorm, wish, work, and review loops backed by markdown plans and local SQLite state.

Developer ToolsClaude Code
Developer Tools Claude Code Security Reviewed
⭐ 323 GitHub stars ⬇ 777/wk npm
INSTALL WITH ANY AGENT
npx skills add agentskillexchange/skills --skill run-genie-planning-execution-loops-in-claude-code Copy
Works best when you want a reusable capability, not another fragile one-off prompt.
At a glance
Tools required
Claude Code, Genie CLI, git, Bun-powered Genie runtime, and a repository where Genie can create .genie markdown documents and a local SQLite state file.
Install & setup
Install from the signed GitHub release or the upstream installer after reviewing provenance; then run genie setup inside the repository and use genie doctor to verify the Claude Code hooks.
Author
Automagik
Publisher
Company
Last updated
Jul 8, 2026
Quick brief

Use this skill when a Claude Code operator wants a structured planning and execution layer before dispatching coding work. The agent should run the Genie setup checks, turn a rough request into a DESIGN.md through brainstorming, convert the design into a scoped WISH.md, dispatch work groups through Claude Code skills, and run an independent review against acceptance criteria before the human sees the final result. The workflow is strongest for repository changes that benefit from explicit planning, task state, worktree fan-out, and a severity-gated review step.\n\nInvoke this instead of using Claude Code normally when a one-sentence request is too ambiguous or too large to hand directly to a coding agent. The boundary is Claude Code-native planning, task decomposition, dispatch, and review, with state kept in markdown and a per-repo SQLite file. It is not a broad agent framework entry; the skill should guide operators through Genie commands and skills such as genie setup, genie doctor, /brainstorm, /wish, /work, /review, genie board, and optional MCP/Omni integrations only when they support that concrete execution loop.