Brigade Stack CLI
The brigadestack CLI keeps shared tasks, memory and verification for several agents: Codex, Claude Code, Cursor and OpenCode.
- 83 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Verification commands run through the CLI, and their output and Git state are stored in the repository.
- Managed execution settings (models, scopes, budgets) define what the agents are allowed to do.
Install
- Works in
Manual install
npm install --global brigadestack@0.30.0The first quick start command from the README installs the `brigadestack` bridge, which downloads a managed Python on first launch.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Verification commands run through the CLI, and their output and Git state are stored in the repository.
- Managed execution settings (models, scopes, budgets) define what the agents are allowed to do.
This is third-party code. Review the repository files before installing.
What it does
Brigade Stack is an open-source coordination layer for agents that work side by side. It sits beside Codex, Claude Code, Cursor, OpenCode and others and keeps state outside any one tool: tasks, dependencies, claims, memory, handoffs and verification records. An agent starts from a work brief (memory, open tasks, earlier evidence), claims a task, runs it within your configured models, scopes, budgets and approvals, and verifies the result through a Brigade command. A verification record keeps the command, exit code, logs and Git state, so another agent can confirm that tests pass. A local setup needs no server; Fleet Hub adds shared ownership and messaging across machines. Two MCP servers exist: memory and skills.
Who it is for. People who run several coding agents on one repository and lose context when tasks pass between them.
Good fit when
- One task goes to Codex, another to Claude Code, and the agents need a shared brief of current work.
- An agent hands work to the next one, which should see blockers and evidence without a chat replay.
- A claim that tests pass has to be recorded with the command, exit code and logs.
- Agents need shared memory with bounded recall and a handoff draft.
Not a fit when
- You work with one agent in one repository: a coordination layer is overkill.
- You need a ready cloud service: the local setup needs no server, and Fleet Hub is configured separately.
- Network requests are forbidden: the Python CLI may send a passive version request.
Example request
Set this repository up for Codex and Claude Code through Brigade, show the dry run, then produce the work brief and record the first verification.Limitations
The stable 0.30.0 version installs from npm and needs Node.js `^22.15.0 || >=23.11.0`; the bridge downloads a managed Python on first launch, and a Python install needs Python 3.11 or newer. Git is required. The main branch is the 0.31 beta line: parallel workers and memory reflection are still in progress, and fleet coordination and governed T3 delegation are only announced. The Python CLI may send a passive version request, and its fields and opt-outs are listed in the telemetry document.
How to disable. Remove the CLI with `npm uninstall --global brigadestack`. Delete files created by `operator quickstart` in the repository and the verification records by hand.
FAQ
How do I install Brigade Stack?
Run `npm install --global brigadestack@0.30.0` and check `brigadestack --version`. The bridge downloads a managed Python on first launch.
How do I record a first verification?
Run `brigadestack work verify run --target . --command "git diff --check" --capture brigade-work`. The record keeps the command, exit code and logs.
Which MCP servers are included?
The memory server starts with `brigadestack memory serve-mcp --stdio`, and the read-only skills server with `brigadestack skills serve-mcp --stdio`.
Does `write_handoff` change memory cards?
No. The tool only drafts a handoff and never edits a canonical card.
Do I need a server?
Not for local work. Fleet Hub adds shared ownership, presence, messaging and model policy across machines.
Discussion
No questions or reviews yet.
Related
CLI toolsA portable .agent folder with memory, skills and protocols that plugs into many coding agents and keeps its knowledge when you switch tools
A bridge between local CLI agents: Codex, Cursor or Claude Code delegate work to Antigravity, Grok, Kimi and seven more.
A memory and coordination layer for coding agents: specs and issues live in git while agents work in parallel
A terminal workspace where several CLI agents like Codex, Claude and Gemini exchange messages and work on one task under human oversight
A shared board for Claude Code, Codex and other agents: a question an agent stops on rises to the top within seconds.
One config for Claude Code and Codex in every repo: shared rules, skill packs, second-model review and a destructive-command guard
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| Brigade Stack (this card) | no | 83 | |
| multi-agents | no | 158 | launches a team of agents and measures whether it beat a single agent; Brigade keeps shared tasks, memory and checks across agents |
| multi-agent-starter | no | 106 | generates a file system for multi-agent work in a folder; Brigade manages tasks and verification records through a CLI and MCP |
README in short
The README explains why a shared layer is needed, shows five working steps, a quick start, a table of compatible tools and the MCP servers, then lists features of the 0.31 beta.