Connect your agent

Building on Tessryx happens through a connected AI agent. Tessryx speaks MCP, the open standard for connecting assistants to tools, so any MCP-capable client works: Claude, or your own.

#The quickest paths (Claude and ChatGPT)

From your dashboard, the Getting started card has two one-click buttons. Add to Claude opens Claude's connector dialog with the Tessryx address already filled in. Try in ChatGPT opens Tessryx's listing in ChatGPT's plugin directory, where Try in chat adds it, with no Developer Mode and nothing to paste. Either way, approve it, authorize your workspace, and you're done. The same card is always reachable from Getting started in the navigation.

#Other clients

Any MCP client can connect using the Tessryx server URL (shown on the Getting started card: copy it from there). Point-and-click setup exists for popular clients; others take a short manual step of pasting that URL into the client's connector settings. The card links out to per-client instructions.

#Authorizing a workspace

When you connect, you approve access on a Tessryx consent screen. Two things to know:

• It's scoped to one workspace. If you belong to more than one, you choose which during authorization, and the connection is pinned to that workspace for its lifetime. To build in a different one, connect again and pick it.
• You're granting build access. Once approved, the agent can read and create resources in that workspace, the same things you can do in the web app. It cannot see your password, and it never sees the values of your secrets.

Two smaller things: if you have two-factor authentication on, approving the connection asks for a one-time code, just like a sensitive account change. And if you don't have a workspace yet, you'll be sent to create one first and returned here to finish.

#What the agent can and can't do

• Can: design schemas, create and edit content, build workflows and pages, wire integrations, set up schedules and domains, everything on the build side.
• Can't: read secret values (you set those), change your account security, or act outside the workspace you authorized.

#Managing the connection

A connected agent shows up as a connected app under Account security. You can revoke it there at any time; the agent loses access immediately, and nothing it already built is affected.

A very busy or looping agent may briefly be rate-limited on its requests. That's isolated to the agent and can never lock you out of the web app.