llms.txt@docs git:20261029.b00419fA

Immutable. This exact content never changes and is served at /api/v1/blob/c7f493a20bfb10bd.

# GraphyLoop

> GraphyLoop is a one-command agentic workflow kit for AI coding agents: it wires Claude Code, OpenCode, Codex, Cursor, Gemini CLI, Oh My Pi and DeepSeek Harness with a 25-agent squad, parallel wave planning, persistent memory and an MCP server. Zero dependencies.

Facts: version 0.6.1 · 7 harnesses · Claude Code plugin · MCP Registry `io.github.chadixearth/graphyloop` · 25-agent squad (26 agent files incl. driver) · 15 slash commands · 15 MCP tools · 70 bundled skills (Agent Skills standard) · zero runtime dependencies · Node.js >= 20 · MIT.

## Start here

- [README](https://raw.githubusercontent.com/chadixearth/graphyloop/main/README.md): positioning, quick start, supported harnesses, how the workflow works, FAQ
- [Landing page](https://chadixearth.github.io/graphyloop/): overview, harness grid, FAQ
- [npm package](https://www.npmjs.com/package/graphyloop): `npx graphyloop`
- [Setup prompt](https://raw.githubusercontent.com/chadixearth/graphyloop/main/docs/SETUP-PROMPT.md): copy-paste prompt that lets any AI harness install and verify GraphyLoop itself

## Docs

- [Harness guide](https://raw.githubusercontent.com/chadixearth/graphyloop/main/docs/harnesses.md): per-harness detection, files written, Claude Code plugin, `~/.agents/skills` for Codex/Cursor, next steps, verification (OpenCode, Claude Code, Codex, Cursor/Windsurf, DeepSeek Harness, Oh My Pi, Gemini CLI)
- [Reference](https://raw.githubusercontent.com/chadixearth/graphyloop/main/docs/reference.md): MCP server internals and benchmarks, configuration, skills, troubleshooting, development, version history
- [Workflow v3 rules](https://raw.githubusercontent.com/chadixearth/graphyloop/main/workflow/AGENTS.md): the harness-neutral 12-rule workflow — recall first/record last, goal + proof, inline first (≤ 8 known files), fan out 2+ independent slices, context pack, frozen contract, turn economy, isolate what collides, no polling/schedule follow-ups, proof not claims, review by blast radius, evidence-first report
- [Changelog](https://raw.githubusercontent.com/chadixearth/graphyloop/main/CHANGELOG.md): release notes
- [Contributing](https://raw.githubusercontent.com/chadixearth/graphyloop/main/CONTRIBUTING.md): ground rules, layout, tests, releasing

## Source

- [GitHub repository](https://github.com/chadixearth/graphyloop): source and issues
- [MCP server and engine](https://github.com/chadixearth/graphyloop/blob/main/lib/mcp.mjs): the 15 MCP tools (plan_feature, task_distribute, memory_search, memory_store, ...)
- [Squad agents](https://github.com/chadixearth/graphyloop/tree/main/agents): the agent prompts
- [Bundled skills](https://github.com/chadixearth/graphyloop/tree/main/skills): 70 skills (Agent Skills standard, installed to `~/.agents/skills` for Codex and Cursor)
- [Claude Code plugin](https://github.com/chadixearth/graphyloop/tree/main/integrations/claude-code): `integrations/claude-code/` plugin (`/plugin marketplace add chadixearth/graphyloop` then `/plugin install graphyloop@graphyloop`)
- [Claude Code marketplace manifest](https://raw.githubusercontent.com/chadixearth/graphyloop/main/.claude-plugin/marketplace.json): `.claude-plugin/marketplace.json`
- [MCP Registry server manifest](https://raw.githubusercontent.com/chadixearth/graphyloop/main/server.json): `server.json` for `io.github.chadixearth/graphyloop` (`npx -y graphyloop mcp`)

## Optional

- [Citation file](https://raw.githubusercontent.com/chadixearth/graphyloop/main/CITATION.cff): how to cite GraphyLoop
- [License](https://raw.githubusercontent.com/chadixearth/graphyloop/main/LICENSE): MIT