What it does
See where agent_ok has been earned. Groups every approval and rejection of an agent-proposed completion by a stored class — a project or a tag — and reports, per class: how many were approved, how many rejected, the current run of consecutive approvals since the last rejection, how much open work there is and how much of it already skips the approval gate. suggestion is "grant" when the streak clears the threshold and there is still gated work to cover, "revoke" the moment a class with granted tasks takes a rejection, and null otherwise. Reports only: it cannot set agent_ok, and neither can you — granting or revoking autonomy is a human action in the app. Use it to tell a human what the record supports, not to act on it.
Parameters
| Name | Type | Required | Description |
|---|---|---|---|
threshold | integer | Optional | Approvals in a row before a grant is suggested (default 5)
|
days | integer | Optional | How far back to read decisions (default 90)
|
Annotations
The hints the server sends with this tool. A host reads them to decide whether to ask you before a call.
- Read-only
readOnlyHint: true - Changes nothing.
- Destructive
destructiveHint: false - Only adds or reads; nothing already there is removed or overwritten.
- Idempotent
idempotentHint: — - Doesn't apply: the tool changes nothing.
- Open world
openWorldHint: false - Touches only your own Done account.
Which tokens can call it
A connection made with OAuth has full access. A token from Settings → Agent access → Extra MCP tokens can be narrowed, and a narrowed token never sees beyond its project or workspace.
- Full accessAllowed
- Read-only tokenAllowed
- Single-project tokenRefused
- Single-workspace tokenRefused
Try it
With Done connected, ask your assistant something like:
“Where have you earned the right to close tasks on your own?”
Connect Done to your assistant
One MCP endpoint per account, over OAuth or a token. Step-by-step setup for each client: