A CLI tool that orchestrates parallel AI coding agents in isolated git worktrees, automating commits, pushes, PRs, and CI monitoring.
Shep is a CLI tool that orchestrates parallel AI coding agent sessions for software development. It creates isolated git worktrees for each feature, runs AI agents to implement code, and automates the subsequent workflow: committing changes, pushing to remote, opening PRs, and monitoring CI. It solves the organizational chaos of managing multiple concurrent AI-driven development tasks.
Developers and engineering teams who regularly use AI coding agents (like Claude Code, Cursor CLI, or Gemini CLI) and want to scale their usage by running multiple feature implementations in parallel without manual branch management and CI oversight.
Developers choose Shep because it automates the entire orchestration layer around AI coding agents—git isolation, commit automation, PR creation, and CI monitoring—enabling true parallel development without conflicts. Its agent-agnostic design and local-first architecture provide flexibility and privacy.
Ship features 10x faster. Burn tokens at scale 😈 (In the right way..)
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
Shep creates separate git worktrees for each feature, enabling multiple AI agents to run concurrently without branch conflicts, as highlighted in the 'Parallel Sessions' section.
From prompt to PR, Shep automates commits, pushes, PR creation, and CI watching with auto-fixes, reducing manual steps for developers.
Works with any terminal-based AI agent like Claude Code or Cursor CLI, allowing users to swap agents per feature without vendor lock-in.
All data is stored locally in SQLite with no cloud services, ensuring privacy and avoiding external tracking, as stated in the '100% Local' feature.
Requires Node.js 22+, Git, GitHub CLI, and pre-authenticated AI agents, with potential sandbox mode conflicts that need manual configuration, as noted in the 'Sandbox mode note'.
Shep only supports specific CLI-based AI agents; if your preferred agent isn't listed or lacks a terminal interface, integration may require custom work.
The CI watch and auto-fix feature relies on clear error messages from CI logs; ambiguous failures can lead to ineffective fixes and manual intervention, as admitted in the FAQ.