# Shep

> Shep is an open-source orchestrator that runs multiple AI coding agents in parallel, each in its own git worktree, from a one-line prompt to a reviewed pull request.

Shep coordinates AI coding agents you already use — Claude Code, Cursor CLI, Gemini CLI, or any agent CLI — so you can ship several features at once without branch switching, stashing, or file conflicts. Each feature gets an isolated git worktree, its own branch, a CI watch loop that auto-fixes failing checks, and a draft pull request; you stay on `main` the whole time. Shep is MIT-licensed, local-first (state lives in `~/.shep/` as SQLite, no Shep servers, no telemetry), and free — a hosted dashboard is also available at app.shep.bot. It is aimed at engineers already using AI coding agents who want to run more than one feature at a time without losing track.

## Docs

- [Docs home](https://shep.bot/docs): Overview, quick start, and a "where to start" table linking every doc.
- [Getting Started](https://shep.bot/docs/getting-started): Install Shep, ship your first feature, and open a draft PR in under five minutes.
- [Core Concepts](https://shep.bot/docs/core-concepts): Index of the four core-concept pages below.
- [How Shep Works](https://shep.bot/docs/core-concepts/how-shep-works): The default flow (describe → agent codes → commit → push → PR → CI watch) and why worktree isolation matters.
- [Features & Tasks](https://shep.bot/docs/core-concepts/features-and-tasks): The feature as the unit of work, its lifecycle fields, and the `shep feat` command family.
- [Parallel Execution & Worktrees](https://shep.bot/docs/core-concepts/parallel-and-worktrees): How git worktrees let Shep run many agents at once without conflicts, and their practical limits.
- [SDLC Lifecycle](https://shep.bot/docs/core-concepts/sdlc-lifecycle): The phases a feature moves through in fast mode vs. spec-driven mode, and the three approval gates.
- [Guides](https://shep.bot/docs/guides): Index of the nine practical guides below.
- [Your First Feature](https://shep.bot/docs/guides/your-first-feature): Step-by-step walkthrough from a blank repo to a merged PR.
- [CLI Reference](https://shep.bot/docs/guides/cli-commands): Every `shep` command and flag, grouped by daemon, features, agents, repos, and settings.
- [Web Dashboard](https://shep.bot/docs/guides/web-dashboard): Every panel, button, and chat input in the localhost:4050 dashboard.
- [CI Watch & Auto-Fix](https://shep.bot/docs/guides/ci-watch): How Shep watches GitHub Actions and retries fixes (default 3 attempts) after a push.
- [Spec-Driven Mode](https://shep.bot/docs/guides/spec-driven-mode): The optional `--no-fast` pipeline with PRD, research, and plan approval gates before code is written.
- [Multi-Repo](https://shep.bot/docs/guides/multi-repo): Running and tracking features across more than one repo from one dashboard.
- [Custom Agents](https://shep.bot/docs/guides/custom-agents): Switching between Claude Code, Cursor CLI, and Gemini CLI, and their capability differences.
- [Configuration](https://shep.bot/docs/guides/configuration): Global settings, per-repo `.shep/config.json`, environment variables, and precedence order.
- [Experimental Features](https://shep.bot/docs/guides/experimental-features): Feature flags for in-development surfaces — Projects, AI Code Review, Agent Collaboration & Supervision, more.
- [FAQ](https://shep.bot/docs/faq): Common questions on workflow, setup, cost, and limitations.
- [Troubleshooting](https://shep.bot/docs/troubleshooting): Fixes for setup, daemon, authentication, feature, CI, and dashboard issues.

## Product

- [Homepage](https://shep.bot): Product overview, install command, feature bento, and idea-to-merged-PR narrative.
- [App dashboard](https://app.shep.bot): Shep's hosted web dashboard.

## Source & community

- [GitHub repo](https://github.com/shep-ai/shep): Source code, issues, and the MIT license.
- [Releases & changelog](https://github.com/shep-ai/shep/releases): Version history and release notes.
- [Discord](https://discord.com/invite/ES6tdVFfur): Questions, bug reports, and feature requests.

## Optional

- [llms-full.txt](https://shep.bot/llms-full.txt): Fuller single-file corpus with expanded core concepts, guide summaries, and FAQ for LLM ingestion.