Skip to main content
AI Agent Orchestration

1 person.1,000 agents.

Ready Agents./Real Work.

Build your AI team. It's easy.

Plan, market, ship, support — every function of your work, run by specialists who remember, collaborate, and answer to you alone.

$npx crewx@latest
Open-source and runs locally on the Claude Code or Codex CLI you already use — nothing new to subscribe to.
Runs on the coding agents you already pay for — CrewX adds roles, workflows, memory, and templates
Codex CLIClaude CodeOpenCodeAntigravity CLICopilot CLI
Live demo

Codex and Claude Code, working together

One request, two coding agents on the job — Codex and Claude Code each take a role and run in parallel, in a single workspace.

Why CrewX

Stop juggling.
Start operating.

From AI tools to an AI team.

A chatbot can answer. A single AI tool can do one task. But real work needs planning, splitting, execution, review, and memory.

CrewX turns the AI tools you already use into a role-based team: a supervisor assigns the work, AI workers execute in parallel, reviewers check the result, and every decision stays in context — across engineering, design, content, support, and analytics.

You supervise the work from the browser. CrewX handles threads, roles, memory, and handoffs so you stop juggling prompts and start operating an AI-native team.

Multi-agent orchestration

Mix Codex, Claude Code, Antigravity, Copilot, and OpenCode in a single crew — Codex handles broad reasoning, Claude Code plans and writes, Antigravity researches, Copilot assists task by task, and OpenCode keeps high-volume work cost-efficient.

Work Instructions (WI)

The PM agent writes structured tickets, then dispatches them to workers. Threaded, traceable, replayable — every change has a paper trail.

Threaded conversations

Each topic lives in its own thread with full history. Agents keep context across days — no more re-pasting your stack.

Long-term memory

Agents save decisions, knowhow, and worklogs as Markdown. BM25-indexed, model-readable, version-controlled with your repo.

Skills & built-in tools

Drop-in skills for memory, search, docs, code review. Compose them per agent. Build your own with a single SKILL.md.

Hooks & automations

Pre/post-tool hooks, scheduled jobs, custom slash commands. Wire CrewX into your existing workflow without leaving the browser.

Feature highlights

It remembers your work,
keeps the process, and lets you undo it.

The parts of CrewX that change how a day of work actually goes — not the ones that look impressive in a demo.

A memory vault for your work

CrewX remembers your work and decisions, so you don't explain the same thing twice.

Screenshot of CrewX's memory vault showing saved project decisions and context

A trace of every step

Every action the AI takes is logged, so you can trace back what happened and why.

Screenshot of CrewX's work trace log showing a timeline of AI actions

Undo, whenever you need to

See exactly what the AI changed, and roll it back if it's not what you wanted.

Screenshot of CrewX's undo panel showing a change ready to be reverted
Your machine
  • · Workspace
  • · Task history
  • · Orchestration
Model providers you choose

You choose, request by request

Runs on your machine

Your workspace, task history, and orchestration run locally on your machine. You choose which model provider sees what, request by request.

More that quietly help

Unified search

Search everything you've ever asked AI, in one query.

Continuity engine

Long tasks keep their context, all the way through.

Redirect mid-task

Talk to the AI again while it's still working.

Checkpoints

Mark a checkpoint, and roll back to it if something goes wrong.

Project rooms

Docs and AI for one job, all in one room.

Instruction loop

Conversations turn into repeatable work processes.

Scheduled runs

AI works on its own, right on schedule.

And more, in the toolkit

Progress you can expand
Preview the prompt before it's sent
Call it up from anywhere
Notices when AI stalls
Notifications
Highlight to flag it
A daily task board
Ask with a photo
Switch workspaces instantly
Coding agents

Claude Code. Codex.
One AI team.

Assign the right coding agent to every role — planning, implementation, review, research. CrewX launches the CLI you are already logged into, so the work runs on the plan you already pay for. Swap any time — no prompt rewrites, no lock-in.

Top pick
Codex
Codex CLI
OpenAI
  • · GPT-5.6
  • · General reasoning
  • · Role: Generalist
Top pick
Claude Code
Claude Code
Anthropic
  • · Opus 5 / Sonnet 5
  • · Long-context planning
  • · Role: Planner & Writer
OpenCode
OpenCode
Community agent
  • · GLM / Qwen / DeepSeek
  • · Low-cost models
  • · Role: Cost saver
Antigravity
Antigravity CLI
Google
  • · Gemini / Claude / GPT-OSS
  • · Web-grounded research
  • · Role: Researcher
GitHub Copilot
Copilot CLI
GitHub
  • · GPT / Claude / Gemini
  • · Pick model per task
  • · Role: Persona simulator

Already have Claude Code or Codex installed? CrewX drives the CLI you are logged into — same account, same quota, no separate CrewX bill.

npx crewx@latest
$ Pricing

Just FREE.

CrewX is free. Bring your own AI subscriptions.

Free

$0

Use CrewX with the AI accounts you already have. No seat fees. No token markup.

Start for free
  • Workspaces20
  • Agents (per WS)100
  • API tokens20
  • Web UI (no install)
  • Bring your own AI subscriptions

Enterprise

Custom

Need higher limits or dedicated support? Let's talk.

Contact us
  • Everything in Free
  • Higher limits & dedicated support

AI provider billing is handled by your own account. Marketplace apps may have their own terms.