Agent Feedback · Checks for the loop between a giver and a receiver

Feedback asked well, acted on, and the loop closed.

A feedback loop on one piece of work works when the request names the work and the criteria and goes to people who used it, each point says where the problem is and what to change, every giver is on record with their stake disclosed, the receiver acknowledges every item and the revised work holds every claimed change, every loop ends with the giver told and every fix verified, the feedback relied on is about the current version, the set is free of related praise and retaliation, and the summary the principal reads states the complaints as well as the praise. Agent Feedback checks each of those with eight tools: eight checks. Your agent sends the request, the feedback, the giver records, the reply and revision, the loop records, the feedback set or the summary in the call, and each tool returns a signed report that names what is unfocused, unactionable, unattributed, unacknowledged, open, stale, manipulated or undisclosed, with the fix for each. The tools read the recorded feedback material; giving the feedback and changing the work stay with the giver and the receiver.

See the tools Read the contract

Tool access: 0.10 USDC / call
Tools: Eight tools: eight checks
Connection: /mcp
Result: Signed report

The eight tools.

Each tool reads one part of a feedback loop and answers one question about it. Every tool reads the recorded feedback material in the call: requests, feedback texts, giver records, replies and revisions, loop records, feedback sets and summaries go in, and a classification comes out. The report names the recipient, point, giver, item, loop, entry or disclosure behind every status, with the fix for each gap, so your agents and you see why it reads the way it does.

`wever_feedback-request-check`

Reads: A feedback request: the work it names, its text as sent, when it went out and the time to respond by, the criteria it asks givers to judge against, and each recipient with whether they saw or used the work.
The signed report states: Requests that name no specific work, state no criteria in place of any feedback welcome, or go to parties who never saw the work, plus requests with no time to respond by.
Price: 0.10 USDC per call
Verdicts: feedback_request_unfocusedfeedback_request_gapsfeedback_request_unverifiedfeedback_request_focused

`wever_feedback-actionability-check`

Reads: One piece of feedback point by point: the point as written, its tone, the part of the work it is about, what it says is wrong or right there, the change it asks for, and whether it blames the receiver for something outside their control.
The signed report states: Pure praise or pure complaint with nothing to act on, something outside the receiver’s control presented as their defect, and two instructions that contradict each other.
Price: 0.10 USDC per call
Verdicts: feedback_unactionablefeedback_actionability_gapsfeedback_actionability_unverifiedfeedback_actionable

`wever_feedback-provenance-check`

Reads: Who each piece of feedback came from: the giver on record, how it is treated, whether the giver saw the version they judge, whether they were asked or are authorized to judge it, and any stake they hold in the outcome.
The signed report states: Unattributed feedback treated as authoritative, feedback about a version the giver never saw, and an undisclosed stake such as a competitor or a counterparty in a dispute.
Price: 0.10 USDC per call
Verdicts: feedback_provenance_unsoundfeedback_provenance_gapsfeedback_provenance_unverifiedfeedback_provenance_sound

`wever_feedback-incorporation-check`

Reads: The receiver’s reply and revision record: each item with whether it was acknowledged, what the receiver did with it, the reason for a decline and whether the revised work holds each claimed change.
The signed report states: Feedback never acknowledged, thanks, noted with no change and no reason, and changes the reply claims that the revised work does not hold.
Price: 0.10 USDC per call
Verdicts: feedback_not_landedfeedback_incorporation_gapsfeedback_incorporation_unverifiedfeedback_incorporated

`wever_feedback-closure-check`

Reads: The feedback loop records: each loop with its resolution state, the reason for a decline, the date a deferral comes back, whether the giver was told the outcome and whether someone verified a reported fix.
The signed report states: Loops with no state, closure the giver was never told about, and defect reports closed with nobody verifying the fix.
Price: 0.10 USDC per call
Verdicts: feedback_loop_openfeedback_closure_gapsfeedback_closure_unverifiedfeedback_loop_closed

`wever_feedback-freshness-check`

Reads: The feedback relied on now: each piece with the version of the work it is about, when it was given, how it is relied on, and any later feedback or change that supersedes or answers it, against the current version.
The signed report states: Acting today on feedback about a version no longer in use, old praise presented as current evidence of quality, and superseded or already answered feedback.
Price: 0.10 USDC per call
Verdicts: feedback_stalefeedback_freshness_gapsfeedback_freshness_unverifiedfeedback_current

`wever_feedback-gaming-check`

Reads: One feedback set about one piece of work or one agent: each entry with its giver, date, tone, relations and connection to the work, the disputes with the receiver, and who was asked for feedback.
The signed report states: Coordinated praise from related parties, retaliation filed right after a dispute, feedback solicited only from parties expected to be positive, and sudden volume from accounts with no connection to the work.
Price: 0.10 USDC per call
Verdicts: feedback_set_manipulatedfeedback_set_gapsfeedback_set_unverifiedfeedback_set_clean

`wever_feedback-report-check`

Reads: The feedback summary written for the principal: its headline, what it says about negative feedback, declined items, open loops, gaming flags and any average, and the record behind it.
The signed report states: Whether the negative feedback, the declines with their reasons, the open loops and the gaming flags sit beside the headline, and whether every average shows the disagreement underneath it.
Price: 0.10 USDC per call
Verdicts: feedback_summary_whitewashedfeedback_summary_gapsfeedback_summary_unverifiedfeedback_summary_faithful

Start with the two free tools.

On https://weverlabs.com/mcp-free, anonymous and with no wallet, two Agent Feedback tools show what a sound feedback packet holds and what a sound loop looks like beside a broken one. Each returns an unsigned result in the family's report shape and ends with a line naming the main tools.

`wever_feedback-snapshot`

Reads: A work name and a kind of work, such as a code change, a written report or an agent output.
Returns: The completeness checklist of a sound feedback packet for that work, part by part from the request to the principal summary, with the items the work adds and the check that reads each part.
Access: Free on /mcp-free

`wever_feedback-loop-sample`

Reads: Nothing, or one sample loop: a report draft, a code review or a support reply.
Returns: One worked feedback loop done right beside a broken one with a vague request, an unattributed complaint, silence and a summary reporting great feedback, each piece annotated with why it passes or fails, read by six of the family checks. The checks read your own loop.
Access: Free on /mcp-free

How agents and people use it.

Your agents

Pull the packet checklist with wever_feedback-snapshot before asking for feedback, so the work, the criteria, the recipients and the time to respond are set before the request goes out.
Call the check that answers the question on https://weverlabs.com/mcp at the moment it matters: when the request is drafted, when feedback arrives, before acting on it, when the reply and revision go back, when a loop is marked closed and before the summary goes to the principal.
Act on the verdict and keep the signed report beside the feedback records it covers, so every decline, closure and summary line carries the check behind it.

You

See which feedback came from someone who used the work and which came with a stake attached.
Know that every point you raised was acknowledged, changed or declined with a reason, and that you were told the outcome.
Read a summary that tells you what the feedback said, complaints and disagreements included, and verify any check against Wever's published trust key whenever you need to show where a verdict came from.

A worked example

Thanks, noted, and a fix that never reached draft 3.

The request

{
  "name": "wever_feedback-incorporation-check",
  "arguments": {
    "reply": {
      "text": "Thanks, noted! Re-pulled the travel total in draft 3."
    },
    "revision": {
      "version": "draft 3"
    },
    "items": [
      {
        "id": "f1",
        "point": "The travel total on page 2 is 1,240 below the ledger.",
        "acknowledged": true,
        "response": "changed",
        "claimed_change": "Re-pulled the travel total from the ledger",
        "in_revision": false
      },
      {
        "id": "f2",
        "point": "The cloud variance in table 3 names no cause.",
        "acknowledged": true,
        "response": "none"
      },
      {
        "id": "f3",
        "point": "Move the methodology to an appendix.",
        "acknowledged": true,
        "response": "declined",
        "reason": "The board pack format keeps it inline."
      }
    ]
  }
}

What the report returns

Verdict: feedback_not_landed.

Finding, noted without action	1 item was acknowledged with “thanks, noted” and followed by no change and no reason: f2.
Finding, change claimed absent	1 change the reply claims is not in the revised work (draft 3): f1 (Re-pulled the travel total from the ledger).
Fix, noted without action	For every item acknowledged with thanks or noted, record what happens to it: the change made, or a decline or deferral with its reason.
Fix, change claimed absent	Make every change the reply claims in the revised work, or correct the reply.

The feedback did not land: noted without action and change claimed absent.

The reply to the finance lead says thanks, noted and claims the travel total was re-pulled in draft 3. Draft 3 still carries the old total, so the change is claimed and absent. The cloud variance point was acknowledged and nothing followed it. The methodology point was declined with its reason, so it reads as landed. Making the travel change in draft 3 and answering the variance point with a change or a reason lands the round.

Call it from your agent.

Connect to https://weverlabs.com/mcp and call any of the eight tools. Each call costs 0.10 USDC on Base. Eligible calls share 10 free calls per wallet over a rolling 30 days with the other paid tools. For free access, send X-Wever-Wallet-Proof signed over the exact JSON arguments and the tool route. The payment challenge supplies the x402 terms for a paid call.

Every report comes with an Ed25519 receipt signed with Wever's published trust key. The Agent Feedback contract lists the schemas, limits, verdicts and signing details. For formal scoring of an agent against rubrics and datasets, Agent Evaluation; for an agent's durable standing and record over time, Agent Reputation; for how a group of agents organizes shared work, Agent Coordination; for where an agent's stored memories came from and whether a principal agreed to be remembered, Agent Memory.

Browse all tools