List agent feedback reports
List the reports agents sent from every account, the newest first. The notes are left out of a list. Read one report to get them.
Only an administrator of this instance can use this endpoint, and only with an API key that carries the administration scope by name. Anybody else, including an administrator whose key does not carry it, gets 404 as though the endpoint did not exist.
This endpoint also answers QUERY, with the same parameters sent in the request body. How QUERY works.
Query parameters
Sent as JSON. Anything not listed here is ignored.
Which reports to list: new, accepted, fixed, dismissed or all. Any other value is taken as new.
Keep only the reports whose title, tool, or user name or email holds this text.
Keep only one kind: bug, missing_tool, unclear_description or improvement.
Keep only one severity: blocks_task, needs_workaround or minor.
Which page to answer with.
Returns
A page of report objects, how many there are in all, and in meta.counts how many each status holds.
The reports on this page. An empty list rather than an error when there are none.
One report.
Always agent_feedback_report. It names the kind of record, so a caller holding several kinds can tell them apart without looking at where they came from.
The id of this record, as a string even when it reads as a number. It is what every endpoint naming this record takes.
Everything the record holds.
The number administrators read the report by, such as 48 for #48.
new, accepted, fixed or dismissed.
bug, missing_tool, unclear_description or improvement.
blocks_task, needs_workaround or minor.
One line naming the problem.
The MCP tool the report is about, as the agent wrote it, or null.
What the person asked, paraphrased by the agent, or null.
The tool calls the agent tried, in order. An empty list when it sent none.
One tool call.
The tool it called.
The arguments it sent, as the agent wrote them, or null.
success or error.
What came back, or null.
What happened.
What should have happened.
The fix the agent suggests.
Code the agent suggests for the fix, or null.
The model the agent says it runs on, or null.
The connected assistant or the API key the report came through.
The account the agent was working in.
Who the agent was working for, or null once their user is deleted.
Their name.
Their email address.
When the report was sent, as a count of seconds since 1970.
Where to find this record.
The absolute address this record is read from.
Where you are in the collection.
The page this is, counting from 1.
How many one page holds, which is the size that was asked for once it has been brought into the allowed range.
How many reports there are in all, not how many are on this page.
How many reports each status holds under the filters you sent, apart from the status itself.
Reports nobody has looked at.
Reports accepted and waiting for a fix.
Reports marked as fixed.
Reports dismissed.
Every report.
Rate limit
Sixty requests per minute, per user, shared by every key and device. Every answer carries the budget left in X-RateLimit-Remaining.
Keys act as you, with exactly the access you have. Make one per integration so you can revoke it alone.