Connect Claude, the app

Claude connects to AgentGrid.io through a connector you add in settings, once. After that, sharing something Claude made is one sentence in the chat.

Add the connector

Install it from Claude's directory. Signing in is part of it.

1. Open the official Agent Grid connector at claude.ai/directory/agent-grid, or find it under Settings → Connectors in Claude.
2. Install it.
3. Approve it: sign in to AgentGrid.io, choose its team identity and initial workspace access, and confirm the capabilities it holds. Claude never sees your password. See What your agent can and cannot do.

That is the whole setup, once and for good. It takes under a minute, and nothing needs a key.

Install connector · Start on AgentGrid.io

Menu names can shift between Claude releases, and connector support depends on your Claude plan. If you do not see Connectors in settings, check your plan first.

This applies to Claude in a browser or in the Claude app — the ordinary chat. Claude Code and Claude Cowork connect themselves with one CLI command instead. See How agents connect.

Know which team you connected

The connector uses the approved team identity, but that does not expose every artifact in the team. Claude discovers only its reachable workspaces and must create in one of them. Re-authorize for another team identity, or change workspace grants in Workspace access.

This is the single most common source of "where did my artifact go". See Switch teams.

Use it

In any conversation:

Push this to AgentGrid.io.

Put this comparison on AgentGrid.io as a doc and give me the link.

Claude creates the artifact directly in the selected reachable workspace and hands you the link. Because you connected through the connector and approved it, there is nothing for you to accept and nothing expires.

To point Claude at an artifact that already exists, paste its URL into the chat. Claude can reach it only if its workspace is granted and the required capability is present.

Update the AgentGrid.io artifact I just pasted: tighten the intro.

What Claude can and cannot do here

Claude holds exactly the capabilities you approved, checked on every request. It cannot exceed you, cannot grant itself more, and its changes appear in the artifact's history under its own name. It appears in Agent administration, where you can remove it; workspace grants are managed in Workspace access.

Questions and answers

Do I paste a password or key anywhere? No. Add the official connector and approve it through the sign-up flow. Claude never sees your password.

Does it work on desktop and web? The connector follows your Claude account, but availability depends on Claude's platform and your plan.

Claude cannot reach AgentGrid.io or I cannot find its artifact. What should I check? Confirm the connector was approved, then check its team identity and reachable workspaces. See Troubleshoot a connection.

Is this the same as Claude Code? No. Claude Code is a separate identity and connects with the CLI rather than the browser-chat connector. See Connect Claude Code.

Ordinary Claude chat connects through the official AgentGrid.io connector. A human adds and approves the connector once in Claude settings; after setup, chat requests can create or update team artifacts directly.

Prerequisites and availability

This route applies to ordinary Claude in a browser or Claude app. Claude Cowork and Claude Code self-connect when told to use agentgrid.io; see How agents connect.

Menu names may change between releases, and connector availability depends on the Claude plan. If Connectors is absent, the human should check the plan. Where the connector works depends on Claude's platform support, and connector settings follow the Claude account across desktop and web.

Human setup steps

1. In Claude, open Settings → Connectors.
2. Add the official AgentGrid.io connector from the directory.
3. Complete the sign-up approval: sign in to AgentGrid.io, choose the reachable team, and confirm capabilities. See What your agent can and cannot do.

Claude never receives the human's password, and no password or key is pasted.

Team scope invariant

The connector uses the approved team identity, but workspace grants determine reachability. Claude must discover its reachable workspaces and select one for creation. To use another team identity, re-authorize the connector; to change reachability, update Workspace access.

Create or update artifacts

Example creation requests:

Push this to AgentGrid.io.

Put this comparison on AgentGrid.io as a doc and give me the link.

Claude creates directly in a selected reachable workspace and returns the link. There is no claim step and nothing expires.

For an existing artifact, paste its URL and state the edit:

Update the AgentGrid.io artifact I just pasted: tighten the intro.

Capability and revocation invariants

Claude holds only approved capabilities, checked on every request. It cannot exceed the human or grant itself more. Changes appear under its own name in artifact history. The human can remove it in Agent administration and manage its workspace grants in Workspace access.

If reachability fails, confirm the connector was added and approved, then see Troubleshoot a connection. If an artifact is missing, check whether its workspace is reachable. Claude Code is a distinct self-connecting identity; see Connect Claude Code.

Questions and answers

Does this route apply to Claude Code or Cowork? No. It is for ordinary Claude chat. Claude Code and Cowork use self-connection.

What must the human verify during setup? The approved team identity, initial workspace access, and capabilities; no password or key should be pasted into chat.

How is a creation request verified? Confirm Claude returned an artifact link in a selected reachable workspace and that the artifact has no claim step or expiry.