parallel agents
Run ten features at once
Every feature gets its own agent session. Launch three, five, ten — your roadmap moves in parallel while you review from one dashboard.
Open-source AI coding agent orchestrator
Run a fleet of coding agents.
Merge real PRs.
Shep runs parallel agents — Claude Code, Cursor, Gemini, or any agent CLI — each in its own git worktree, through implementation, CI, and review. Open source, local-first, cross-platform.
npx @shepai/cli
Runs locally with Node 22+. Opens the dashboard at localhost:4050.
Free and MIT-licensed. Bring the agent subscription you already have.
app.shep.bot
idea → spec → agents → review → merged PR
The whole loop, not just the coding.
Every feature moves through the same pipeline — from a sentence to a merged pull request.
- Describe
One sentence in, one branch out.
Type what you want in the dashboard or run shep feat new. Shep cuts a branch and a worktree.
- Spec
Optional gates for complex work.
Enable spec-driven mode and Shep writes requirements, research, and a plan as YAML you approve before any code.
# specs/003-add-pagination/spec.yaml name: add-pagination number: 003 oneLiner: Cursor pagination for GET /todos. requirements: - page + limit query params, default limit 20 - response includes total count tasks: - id: task-1 # cursor decode - id: task-2 # paginated query gate: awaiting your approval
- Run agents
Parallel, isolated, yours.
Each feature gets its own git worktree and agent session — Claude Code, Cursor, Gemini, or any agent CLI.
- Review
You hold the merge button.
Shep commits, pushes, opens a draft PR, watches CI, and pushes fixes when it fails. You review the diff.
- Merge
Real PRs, your process.
Features land as standard pull requests on named branches. Your CI, your reviewers, your merge — unless you flag --allow-merge.
Openfeat: cursor pagination for GET /todosfeat/add-paginationmain
- lintpassed
- typecheckpassed
- test (unit)passed
- test (int)passed
- buildpassed
all checks passed — awaiting your review
features
Everything you need to ship on autopilot
worktree isolation
A worktree per feature
Each AI agent works in its own git worktree on its own branch. No clobbered files, no port collisions, no stash juggling. Your checkout stays untouched.
CI watch
CI watch and auto-fix
After push, Shep watches your pipeline. When it fails, the agent reads the logs and pushes a fix — up to three retries, configurable.
spec-driven
Specs and approval gates
For complex features, flag --no-fast: Shep drafts requirements, research, and a plan as YAML, then pauses at gates you approve.
agent-agnostic
Bring your own agent
Claude Code, Cursor CLI, Gemini CLI — or any agent that runs in a terminal. Swap per feature, per repo, anytime. No vendor lock-in.
web + CLI
Dashboard and CLI, equal
A live graph of every repo and feature at localhost:4050 — status, diffs, chat. Prefer the terminal? Every action is a shep command.
open source
Open source, local-first
MIT-licensed. All state lives in ~/.shep/ as SQLite. Nothing goes to Shep servers — there are none.
what is shep
The open-source dashboard for parallel AI coding agents
Shep is an AI coding agent orchestrator that lets you run multiple Claude Code, Cursor, and Gemini sessions in parallel — each in its own git worktree, from prompt all the way to pull request. Manage many feature builds at once from a single local dashboard.
faq
Frequently asked questions
Run `shep feat new` once per feature. Shep gives each session its own git worktree, branch, and terminal, so parallel Claude Code sessions never touch the same files. Shep manages the Claude Code worktrees for you — creating them, committing in them, and opening PRs from them — while you monitor everything from one local AI coding dashboard. The same goes for Cursor CLI and Gemini CLI.
A control plane that runs and coordinates multiple AI coding agents: it assigns each one an isolated workspace, drives the git workflow around it — branch, commit, push, PR — watches CI, and reports status in one place. Shep is an open-source AI coding orchestrator: MIT-licensed, local-first, and agent-agnostic, so one dashboard manages parallel AI coding agents from different vendors.
Every feature gets a dedicated git worktree on its own branch. AI agents working on different features never share a checkout, so there are no clobbered files, port collisions, or half-finished edits crossing between tasks — and your own working directory is never modified. When the agent finishes, Shep commits, pushes, and opens the pull request from that worktree, which is preserved until you delete it.
Claude Code, Cursor CLI, and Gemini CLI today. Shep is agent-agnostic by design — agents run through a generic executor interface, so if an agent runs in a terminal, Shep can orchestrate it. Swap agents per feature or per repo: run Claude Code on one feature and Gemini CLI on another at the same time.
Yes. Shep is MIT-licensed and runs entirely on your machine. All state lives in `~/.shep/` as a SQLite database, and Shep makes no calls to Shep servers — there are none. Your code goes only to the agent you configure, under that agent's own terms. Start with `npx @shepai/cli` — no account required.
Shep's prompt to PR loop puts three layers between the agent and your main branch: the agent works in an isolated worktree, never your checkout; its output lands as a draft pull request that your CI, linters, and security scanners run against; and nothing merges without your approval unless you explicitly pass `--allow-merge`. Every action is logged, and `shep agent stop` halts a run immediately — the worktree is preserved.
Free and MIT-licensed
Bring the agent subscription you already have.
Shep adds the workflow around your agent — worktrees, commits, CI watching, pull requests. No Shep account, no usage meter, no servers on our side. The source is on GitHub; read it before you run it.
npx @shepai/cli
View on GitHub
Live count, straight from the API. We don't cache our own traction.
Open source
local-first · no Shep servers
Works with Claude Code, Cursor CLI, and Gemini CLI — and any agent you can launch from a terminal.