AGENTS & API

A multisig your agents can work. You still hold the keys.

Give Claude, Cursor, a CI job or your own bot native access to the request queue. They read, propose, review and co-sign. The contract still enforces the quorum, so the tools work the same for people and for agents.
Create an API keyRead the setup guide

The safety model

Keys are scoped

read, propose, sign, manage. Restrict a key to specific wallets, give it an expiry, revoke it instantly. A key can never sign a transaction or mint another key.

Signing stays local

To approve or execute, the agent uses its own owner key on its own machine. It is never sent to mymultisig.app.

A human in every decision

Pick a threshold the agent can never meet alone. In a 2-of-3 with two humans plus the agent, it can prepare and co-sign, but nothing moves without a human signature.

Four ways in, one set of credentials

Every flow below uses the scoped API keys you create in Settings → API keys. Approvals and executions are signed locally with the agent's own owner key — never on our servers.

mcpalpha

MCP server

Give Claude, Cursor or any MCP client native multisig tools.

  • Read wallets, balances and the request queue
  • Decode requests with risk warnings and locally verified signatures
  • Propose, approve and execute (simulated first) with the agent’s owner key
  • Read-only mode and a chain allowlist for safe rollouts

npx -y mymultisig-mcp — add it to any MCP client config with your API key.

Setup guidenpm
skillsalpha

Claude Code skills

Ready-made multisig workflows for Claude Code, MCP server included.

  • multisig-status: owners, balances and what the queue needs next
  • create-multisig and propose-transaction with sanity checks
  • review-and-sign: decode, flag risks, sign only after you confirm
  • execute-queue: run ready requests in nonce order

/plugin marketplace add marc-aurele-besner/mymultisig-app, then /plugin install mymultisig@mymultisig.

Setup guideSource
apialpha

REST API & API keys

Call the request queue and wallet registry from any backend or bot.

  • Scoped Bearer keys: read, propose, sign, manage
  • Restrict a key to specific wallets and give it an expiry
  • Revoke instantly; keys can never sign transactions or mint keys
  • OpenAPI 3.1 spec and llms.txt for code generators and LLMs

Keys act as the wallet that created them — the same key works for the MCP server and the CLI.

Create an API keyOpenAPI spec
clialpha

Command line

Script your multisig from a terminal, CI job or agent shell.

  • Create wallets, propose, approve and execute requests
  • Human-readable output, or --json for scripts and agents
  • Irreversible actions ask first, or need --yes when unattended
  • Mint and revoke API keys by signing in with your key

npx mymultisig-cli --help

Setup guidenpm

Set up in four steps

  1. 01Create an API keyMint a scoped key while signed in with your wallet.Open API keys →
  2. 02Give the agent an owner key (optional)Generate a fresh key and add its address as an owner so it can sign.
  3. 03Connect the MCP serverAdd mymultisig-mcp to Claude Code, Claude Desktop, Cursor or any MCP client.Copy the setup →
  4. 04Ask the agentCheck status, propose, review, sign and execute through natural language.

Build your own

Everything the app does is public. If you need an integration we don't have, the pieces are already on the table.

contractsABIs on npm

Factory and wallet ABIs ship in the mymultisig-contract package.

onchainStandard EVM calls

Every read and write is a plain contract call — wagmi, viem, ethers all work.

appOpen API routes

The app’s request storage and signing routes are open source.

sdkTypeScript SDK

mymultisig-sdk: the same EIP-712, signature and execution code this app runs, on viem.

Want requests in Slack, Discord, Telegram or email instead?

Team chat & notifications