Report Wrong Data - Sixtyfour AI Documentation

Send POST /feedback with a rating, a message, or field verdicts. Include the task or workflow row when you have it; leave out target to report a problem with the API, docs, or a tool.

Send a report

This example reports a problem with a People Intelligence response. Replace the message with what happened. Reuse the same Idempotency-Key when retrying this report; use a new key for a different report.

curl --fail-with-body https://api.sixtyfour.ai/feedback \
  -H 'X-API-Key: YOUR_API_KEY' \
  -H 'Content-Type: application/json' \
  -H 'Idempotency-Key: email-review-001' \
  -d '{
    "category": "outdated_data",
    "message": "The returned email bounces. The person has changed employers.",
    "operation": "POST /people-intelligence",
    "source": {"kind": "agent", "agent_name": "lead-reviewer"}
  }'

import json
from urllib.request import Request, urlopen

request = Request(
    "https://api.sixtyfour.ai/feedback",
    data=json.dumps({
        "category": "outdated_data",
        "message": "The returned email bounces. The person has changed employers.",
        "operation": "POST /people-intelligence",
        "source": {"kind": "agent", "agent_name": "lead-reviewer"},
    }).encode(),
    headers={
        "X-API-Key": "YOUR_API_KEY",
        "Content-Type": "application/json",
        "Idempotency-Key": "email-review-001",
    },
)
with urlopen(request) as response:
    print(json.load(response))

const response = await fetch("https://api.sixtyfour.ai/feedback", {
  method: "POST",
  headers: {
    "X-API-Key": "YOUR_API_KEY",
    "Content-Type": "application/json",
    "Idempotency-Key": "email-review-001",
  },
  body: JSON.stringify({
    category: "outdated_data",
    message: "The returned email bounces. The person has changed employers.",
    operation: "POST /people-intelligence",
    source: { kind: "agent", agent_name: "lead-reviewer" },
  }),
});
if (!response.ok) throw new Error(await response.text());
console.log(await response.json());

The receipt includes the ID to keep for checking status:

{
  "id": "6a0b5d0e-8f1e-4c2a-b7d3-9e8f7a6b5c4d",
  "already_submitted": false,
  "target": null,
  "fields_recorded": 0,
  "created_at": "2026-11-05T18:22:11Z"
}

Identify the result and correction

Add target and fields to the request when you can identify the result:

{
  "target": {"type": "task", "task_id": "enrich_7d1c0e52-5b8f-4c1a-9f0e-3a2b1c4d5e6f"},
  "fields": [{
    "path": "email",
    "verdict": "outdated",
    "returned": "jane@acme.com",
    "correct": "jane@newco.com",
    "evidence_urls": ["https://www.linkedin.com/in/janedoe"]
  }]
}

Replace the example task ID with the task_id from your async response or webhook. For a synchronous enrichment, use its X-Correlation-ID response header. If a request failed before creating a task, leave out target and include its correlation_id in message. For a workflow row, use {"type":"workflow_row","run_id":"YOUR_JOB_ID","block":2,"row":37}: run_id is the job_id from POST /workflows/run, block is the results file’s block_number, and row is its _row_id. Use wrong for an incorrect value, outdated for a value that was right once, missed when a value exists but we returned nothing, or correct to confirm a value. Leave unreviewed fields out. Leave out correct when you don’t know the right value; an explicit null reports that the right value is empty.

Check review status

Use the receipt’s id with the same team credentials:

curl --fail-with-body https://api.sixtyfour.ai/feedback/6a0b5d0e-8f1e-4c2a-b7d3-9e8f7a6b5c4d \
  -H 'X-API-Key: YOUR_API_KEY'

The response includes your report, its channel, status, and status_updated_at. Status starts at received, changes to reviewed when the team has examined it, and to fixed when the reported issue is resolved. A status change does not rewrite a result you already downloaded. Access to feedback follows access to its target.

From an MCP agent

Call sixtyfour_send_feedback with the same report in feedback and a stable idempotency_key. Then call sixtyfour_get_feedback with the returned ID to check its status. Ask your agent: “Report this incorrect email to Sixtyfour, include the task ID and evidence, and save the feedback ID.” See the MCP setup guide to connect these tools.

Errors

For error responses (400, 403, etc.), see Handling Errors.

​Send a report

​Identify the result and correction

​Check review status

​From an MCP agent

​Errors

Send a report

Identify the result and correction

Check review status

From an MCP agent

Errors