The control plane for coding agents

Every agent. Every machine. Landed.

AGI CLI runs Claude Code, Codex, Cursor, Grok and eleven more harnesses across your laptop and your machines, keeps them moving when they stall, and drives each task to a merged pull request. On the subscriptions you already pay for.

$curl -fsSL agi-cli.sh | sh

or npm i -g @phnx-labs/agents-cli. Lands on your PATH as agents and ag. macOS and Linux.

Claude Code

Codex

Cursor

OpenCode

OpenClaw

CopilotGrokKimiAntigravityDroid

GooseAmpWarpHermesMuse

Starting an agent is the easy part. Agents stall. They stop mid-task, ask a question and idle, or hand the work back. AGI CLI notices the one that stopped and gets it moving, so the work lands without you watching every session.

From prompt to pull request

One task, three machines, one merge

checkoutmerged

$ agents teams create checkout --devices mac-mini,linux-1,linux-2
$ agents teams add checkout claude "Plan the rewrite" --name plan
$ agents teams add checkout codex "Owns api/*" --name api --after plan
$ agents teams add checkout cursor "Owns app/*" --name ui --after plan
$ agents teams add checkout claude "Run Playwright" --name qa --after api,ui
$ agents teams start checkout --watch
  plan   done      mac-mini   4m 12s
  api    running   linux-1
  ui     running   linux-2
  ui     idle 4m   asked: "keep the legacy coupon field?"
  watchdog  nudged ui to continue
  api ui   done
  qa     done      mac-mini   38 checks passed
  checkout  pull request merged into main

The command surface

Seven verbs that cover the day

Run

One entry point for every harness. Pin versions per repo, spread runs across your accounts, or put a run on another machine.

Docs

Coordinate

Teammates with owners and an order. Independent branches keep going when one fails; only its dependents wait.

Docs

Configure

Skills, commands, hooks, MCP servers and rules live once in ~/.agents/ and sync into each agent's own format.

Docs

Steer

See what every agent is doing, message one mid-task, and let the watchdog nudge the ones that stalled.

Docs

Dispatch

Hand a task to a cloud provider and get a pull request back, or keep it on your own boxes.

Docs

Inspect

Every transcript from every agent, indexed and searchable across machines. Then read what the runs say about how you work.

Docs

Schedule

Routines put repeatable agent work on a cron: standups, nightly audits, PR triage. One daemon fires each one once.

Docs

runzsh

$ agents run claude "fix the flaky checkout test"
$ agents run codex --balanced "run the full suite"
$ agents run auto --device auto "upgrade the logger"

Devices

Your machines are one pool

Every box on your Tailscale network becomes a place an agent can run. agents run --device auto picks the least-loaded worker, teams spread teammates across the pool, and transcripts sync back so a session is searchable and resumable from any machine. Everything travels over SSH on your own network.

Point at a machine to see what it is running.

What you are trusting

Free, local, and yours to read

15 harnesses

Claude Code, Codex, Cursor, Grok, Kimi, Droid and more, through one command shape.

$0 per token

Runs on the subscriptions and API keys you already have. No account required.

0 usage data

The CLI sends no telemetry. Its one outbound call is a daily version check against npm.

SSH transport

Machines reach each other over SSH on your own network. No relay or hosted service in between.

1 config

~/.agents/ is the source. Every harness gets it in its own format, on every machine.

Compared

Against every other way to run agents

Most tools give you one piece. agents gives you all of them.

Capability	agi-cli	Claude Code alone	Cursor	Run CLIs by hand	OpenCode
Pin versions per project (.nvmrc-style)	✓	—	—	—	—
Run multiple agents (Claude + Codex + Grok) from one CLI	✓	—	—	~	—
Swap underlying model (Kimi, GLM, DeepSeek via OpenRouter)	✓	—	~	—	✓
Rotate across multiple accounts to dodge rate limits	✓	—	—	—	—
Parallel teams with DAG dependencies	✓	—	—	—	—
Distribute teammates across machines over SSH	✓	—	—	~	—
Steer & auto-nudge running agents mid-flight	✓	—	—	—	—
Managed cloud run that opens a PR	✓	—	~	—	—
Reusable multi-agent workflow bundles	✓	—	—	—	—
Local browser via CDP (drive any site)	✓	—	—	—	—
Cross-agent session search & replay	✓	~	—	—	~
Cron / scheduled routines	✓	—	—	~	—
Keychain-backed secrets (no .env files)	✓	—	—	—	—
Sync skills/MCP/commands across all installed agents	✓	~	—	—	~
Source-available and self-hosted — cloud dispatch optional	✓	—	—	✓	✓

Pin versions per project (.nvmrc-style)

agi-cli✓
Claude Code alone—
Cursor—
Run CLIs by hand—
OpenCode—

Run multiple agents (Claude + Codex + Grok) from one CLI

agi-cli✓
Claude Code alone—
Cursor—
Run CLIs by hand~
OpenCode—

Swap underlying model (Kimi, GLM, DeepSeek via OpenRouter)

agi-cli✓
Claude Code alone—
Cursor~
Run CLIs by hand—
OpenCode✓

Rotate across multiple accounts to dodge rate limits

agi-cli✓
Claude Code alone—
Cursor—
Run CLIs by hand—
OpenCode—

Parallel teams with DAG dependencies

agi-cli✓
Claude Code alone—
Cursor—
Run CLIs by hand—
OpenCode—

Distribute teammates across machines over SSH

agi-cli✓
Claude Code alone—
Cursor—
Run CLIs by hand~
OpenCode—

Steer & auto-nudge running agents mid-flight

agi-cli✓
Claude Code alone—
Cursor—
Run CLIs by hand—
OpenCode—

Managed cloud run that opens a PR

agi-cli✓
Claude Code alone—
Cursor~
Run CLIs by hand—
OpenCode—

Reusable multi-agent workflow bundles

agi-cli✓
Claude Code alone—
Cursor—
Run CLIs by hand—
OpenCode—

Local browser via CDP (drive any site)

agi-cli✓
Claude Code alone—
Cursor—
Run CLIs by hand—
OpenCode—

Cross-agent session search & replay

agi-cli✓
Claude Code alone~
Cursor—
Run CLIs by hand—
OpenCode~

Cron / scheduled routines

agi-cli✓
Claude Code alone—
Cursor—
Run CLIs by hand~
OpenCode—

Keychain-backed secrets (no .env files)

agi-cli✓
Claude Code alone—
Cursor—
Run CLIs by hand—
OpenCode—

Sync skills/MCP/commands across all installed agents

agi-cli✓
Claude Code alone~
Cursor—
Run CLIs by hand—
OpenCode~

Source-available and self-hosted — cloud dispatch optional

agi-cli✓
Claude Code alone—
Cursor—
Run CLIs by hand✓
OpenCode✓

In depth: Factory AI, OpenCode, Cursor, CLIs by hand

Questions

Before you install

What is agi-cli?

agi-cli (AGI CLI) is a source-available meta-harness that runs every CLI coding agent — Claude Code, Codex, Cursor, Grok, and more — from one interface. Install it with `curl -fsSL agi-cli.sh | sh`; it lands on your PATH as `agents` (alias `ag`). One install, one config, every agent.

Is agi-cli free?

Yes — every feature, for every user, forever. It runs agents on your existing subscriptions, so there are no per-token API costs and no account required. An optional sign-in adds team spaces; nothing is gated behind it.

What license is agi-cli under?

FSL-1.1-Apache-2.0. Use it, modify it, redistribute it — for any user or company, free of charge. The only thing barred is offering agi-cli — or a substitute for it or for other Phoenix Labs products built with it — as a competing commercial product or service. Each version converts to plain Apache-2.0 two years after it ships.

Which coding agents does agi-cli support?

Claude Code, Codex, Cursor, OpenCode, Grok, Kimi, Droid, Antigravity and more — plus any model (Kimi, MiniMax, GLM, Qwen, DeepSeek) through OpenRouter profiles. You run them individually, pin versions per project, or spawn them as parallel teams.

How is agi-cli different from Claude Code?

Claude Code is one agent. agi-cli runs and orchestrates all of them — version-pinned per project like .nvmrc, in parallel teams with boundary contracts, with skills, MCP servers, secrets, and searchable sessions shared across every agent from one config source.

How do I install agi-cli?

Run `curl -fsSL agi-cli.sh | sh`. It runs on macOS and Linux.

Install it, then dispatch something

$ curl -fsSL agi-cli.sh | sh
$ agents run claude "read this repo and tell me what is broken"

Read the quickstart