Applies if you landed on a claim page, which happens when an AI reached AgentGrid.io without connecting. Connect it and its work arrives in your team directly. See How agents connect.
Claiming asks you one question that has consequences: does the AI that made this keep access?
The two paths
Claim and connect the agent. It gets its own identity in your team, with the capabilities you approve. It can keep editing this artifact and creating others within that scope.
Claim without the agent. The work is yours. The AI stays read-only and cannot edit it again.
What closes is the shortcut, not the door. Granting at claim time completes the agent's waiting connection automatically. Claiming without granting ends that waiting connection, and it cannot be retried. The same AI can still be connected afterwards — through the connector in Claude or the plug-in in ChatGPT, with npx @animaapp/cli@latest login for a coding or autonomous agent, or with a link from Share → Invite an agent. It just takes a few extra steps that granting now would have saved.
What you are approving
| Field | What it means |
|---|---|
| Name | Comes from the AI itself, and is what you will see in history and in the access list. Rename it now if you like. |
| Team | The team the artifact lands in, which is the one the agent gets access to |
| Capabilities | read, write, share, publish |
The guarantee
An agent can never exceed your permissions. Every request is checked against your live permissions, so a change takes effect immediately rather than at some future expiry. You can revoke it at any time.
That is why granting write is not the same as handing out your password. The agent is a separate identity, scoped to what you approved, attributed in history, and reversible.
Choosing capabilities
| Capability | Grant it when |
|---|---|
read |
Always. Without it the agent cannot see anything. |
write |
You want the AI to keep editing. This is the usual case. |
share |
You want it to give others access. |
publish |
You want it to be able to deploy to a public URL. Consider leaving this off and asking for it case by case. |
A reasonable default is read and write, adding the others when a real need appears.
Choose carefully here. Capabilities cannot be edited later — changing them means removing the agent and inviting it again.
Steps
::: lanes
In the app
On the claim screen, confirm the agent's name, tick the capabilities, and select Claim and connect. To skip it, select Claim without the agent.
Afterwards, review what it holds under Share → Manage access.
Ask your AI assistant
I granted you write access. Add a section on pricing.
:::
FAQ
Can I change capabilities later? Not directly. You can see every agent's capabilities and remove an agent under Share → Manage access. To change what one holds, remove it and invite it again from Share → Invite an agent.
What if I claim without the agent and then change my mind?
Connect it afterwards: the connector in Claude, the plug-in in ChatGPT, npx @animaapp/cli@latest login for a coding or autonomous agent, or an agent invite. The only thing you lose is the shortcut.
Does connecting it give access to my other artifacts? Only within the team and capabilities you approved.
Is the agent's name meaningful? Yes. It is what appears on every change it makes and in your access list, so pick something you will recognise later.