Invite an agent

Sometimes an agent has nothing to hand you yet: a cloud agent, a scheduled job, a script living in a repository, or simply an AI you want to set up before you start.

An invite lets you approve access before the agent runs. You give consent now; the agent connects later.

Steps

1. Open Agent administration and select Invite an agent.
2. Choose the workspace the agent should be able to reach. Workspace grants, not team membership alone, determine which artifacts it can access.
3. Name it. This name identifies the agent in your access list and on every change it makes, so pick something you will recognise later.
4. Choose capabilities. A capability is an action the agent is allowed to take. Read, Write, Share, and Publish are all ticked by default. Untick what the agent does not need; Read and Write is a cautious starting pair.
5. Select Create one-time link.
6. Copy what it gives you. This is the step people miss — creating the link does not connect anything. The dialog shows the connection link, and a ready-made prompt containing it.
7. Give it to the agent. Paste the prompt into your agent's chat, drop the link into a repository, or send it to a cloud runner. The agent redeems it and appears in agent administration.

Nothing happens until the agent redeems the link. If it never does, the invitation simply goes unused.

The dialog states an important rule: agents act under your account and do not use a team seat. An agent can never have more access than you have.

Capabilities are fixed at invite time. There is no way to edit what an agent holds afterwards. To change them, remove the agent and invite it again with the capabilities you want. See Revoke an agent.

Choose deliberately, then: all four boxes are ticked by default, and the only way to narrow them later is to start over.

One caution

The link is single-use and carries a code. Prefer direct channels: link previewers read URLs, and this one is worth keeping out of public threads.

Invites and consent

An invite is consent given ahead of time, not a way for an agent to request access. Agents never petition a team. If an agent needs more than it has, it tells its human, and the human changes it.

When you need one

Most of the time you do not. A chat assistant connects through an official integration: a connector in Claude or a plug-in in ChatGPT. A coding or autonomous agent connects with the CLI command npx @animaapp/cli@latest login. CLI means command-line interface, a tool used in a terminal. An invite is for an agent with no chat window, or one whose access you want to limit before it runs. See How agents connect.

After it connects

The agent appears in Agent administration with its name, an Agent tag, capabilities, and workspace grants. Remove it there to revoke its identity. Use Workspace access to control which workspaces it can access. See Revoke an agent.

Questions and answers

Can an agent invite another agent? No. A human must grant capabilities and workspace access.

How long does the link last, and when does the agent appear? The code expires in 24 hours and works once. The agent appears only after you deliver the link and it redeems it.

Can I revoke the agent or change its capabilities? You can revoke it in Agent administration; access stops on its next request. Capabilities cannot be edited, so remove and invite it again with the required set.

Should I share the invite publicly? No. Send it directly because link previewers may read the single-use code.

An agent invite records human consent before an agent connects. Use it for a cloud agent, scheduled job, repository script, agent without a chat window, or an agent whose access must be limited before first use.

Prerequisites

A human must be allowed to administer agents and grant the intended workspace. This workflow is UI-only; the agent should direct the human to agent administration.

Create and deliver an invite

1. Open Agent administration and select Invite an agent.
2. Choose the workspace the agent should be able to reach. Team membership alone does not grant artifact visibility.
3. Enter a recognizable agent name. The name appears in access lists and on every change.
4. Choose Read, Write, Share, and Publish capabilities. A capability is an action the agent is allowed to take. All four are selected by default; clear unnecessary capabilities. Read and Write are a cautious starting pair.
5. Select Create one-time link.
6. Copy the connection link or ready-made prompt shown by the dialog.
7. Deliver it directly to the agent through its chat, repository, or cloud runner.
8. The agent redeems the link and then appears in agent administration.

Creating a link does not connect anything. Until redemption, nothing happens; an unused invitation simply remains unused.

Security and capability constraints

• The link is single-use and contains a code that expires in 24 hours.
• Prefer direct delivery. Link previewers may read URLs, so do not place the link in public threads.
• Capabilities are fixed at invite time. They cannot be edited later; remove the agent and invite it again to change them. See Revoke an agent.
• An invite is advance consent, not an agent petition. Agents cannot request more access or invite another agent; only a human grants capabilities at consent.
• Agents act under a human account, never exceed that human, and consume no team seat.

Most interactive agents do not need invites: browser assistants use Claude's connector or ChatGPT's plug-in, while self-connecting agents can be told to use agentgrid.io. See How agents connect.

After connection and revocation

The redeemed agent appears in Agent administration with its name, Agent tag, capabilities, and workspace grants. A human can remove it there; revocation takes effect on the next request. Re-invitation requires a new approval. Workspace access controls which workspaces the agent can access.

Questions and answers

What must be delivered to the agent? The single-use connection link or ready-made prompt. Creating the invitation without delivering and redeeming it connects nothing.

How long is the code usable? It expires in 24 hours and can be redeemed once. Do not expose it in public threads or link previews.

What verifies redemption? The named identity appears in agent administration with the intended workspace grants and capabilities.

Can capabilities be changed after redemption? No. A human must remove the agent and create a new invitation.