Forculus Start free

Agents

Agents with names of their own

On Forculus an AI agent, a nightly job or another service can be an agent of your workspace: it has its own name, reaches only the apps you grant it, and calls them with credentials you can revoke. Every request it makes is in the audit log under its name, and revoking it closes its open connections within 15 seconds. Agents are on every plan.

Updated

Make one and grant it an app

forculus agent create acme release-bot
forculus agent grant acme/payroll release-bot --role viewer
forculus agent token acme release-bot

The credential starts with fca_ and is shown once. We keep only a hash of it. Give it to the agent in its secret settings, never in code: a deploy with one in its files is refused. A credential always ends, after up to 365 days.

It reaches no further than its maker

An agent gets into an app only through a grant. Being in the workspace, or an app being open to the whole workspace, lets people in, never agents. An agent made by a member never reaches further than that member can: take their access away and their agents lose it at the same time.

What your app sees

Each request carries a fresh signed identity token. For an agent it says it's a service, names the agent and gives its role on the app. Our SDKs for Node, Python, Go and Rust tell an agent from a person in one call. See What your app sees.

Stop it

Revoke a grant or a credential and the agent's next request is refused. An open WebSocket, stream or download closes within 15 seconds. Ending an agent is for good: its credentials stop, its apps are taken away, and the log keeps what it did.

forculus agent end acme release-bot

Coding agents act as you

A coding agent that you drive, such as Claude Code, Codex or Cursor, deploys as the person signed in to the CLI and shows under their name. An agent of the workspace is for software that runs on its own. See Claude Code.

Questions

How is an agent different from a bypass token?

A bypass token opens one app and stops when the person who made it loses access. An agent can hold several apps, outlives the person who made it, and is managed by the workspace.

How many agents can a workspace have?

100 at once, each with up to 10 live credentials.

Can an agent call an MCP server deployed here?

Yes, once the app is granted to it. People connect AI clients to an MCP server the same way, as themselves. See MCP servers.

Next