AI agents & MCP — Willo docs

Connect Claude Code, Cursor or Codex CLI to your business: tokens, drafts, approvals, limits and security.

Get started

Connect your agent: Connect Claude Code, Cursor or Codex CLI to your Willo business in about five minutes, with a token you create on the dashboard.
Your first change: Walk your agent through one safe change to your site, from reading a file to seeing it on the draft preview and publishing it.

Draft and live: How Willo keeps a private draft of your site and backend apart from the live site, and what that means for changes your agent makes.
Approvals: Which actions your agent can't take without you, how an approval card reaches you, and how the agent waits and retries after you decide.
Agent tokens: What an agent token can and can't do, how read and act scopes follow the creator's role, expiry, revoking and rotating, and keeping one safe.
Limits and billing: What your agent's calls cost, the rate limits per token and per business, and what to do when a call is told to slow down.
Security model: How Willo keeps an agent inside the same rules as its own chat, refuses secrets, protects personal data, and records everything an agent does.
Handing Willo a task: When your agent should give Willo's chat a whole job with willo_task instead of calling tools itself, what it costs, and how to read the result.

Deploy an edge function from your agent: Have your agent create, change and redeploy one of your backend's edge functions, including functions split across several files.
Change the database schema from your agent: Add tables and columns to your backend from your agent, when a change runs at once and when it needs a card, and how drafts protect live data.
Edit CMS content and publish: Have your agent write and update blog posts and other CMS items as drafts, then publish, schedule or unpublish them with your approval.