AI API Response → Potential Issues

AI Bug Analysis

A second pair of eyes on API output — the AI reviews structure, values and headers for defects, security exposures and contract smells, with severity and suggested fixes.

Privacy: Requires a free Mutqan account. Your input is sent to the Mutqan AI service to generate the result; prompts and results are not stored — only your account's usage counts for the daily allowance. How the AI tools work

What does a good API error response look like?

A good API error uses the right status code and a consistent, machine-readable body. RFC 9457 (Problem Details) standardises that body: a type URI, a short title, the status, a human-readable detail and an instance identifier, in application/problem+json. It tells the caller what went wrong and how to fix it without exposing stack traces, SQL or internal paths.

How to use AI API Response → Potential Issues

  1. Paste the response — the status line, headers and body, or just the JSON.
  2. Optionally paste the expected contract (a schema or a description) and the request it answered.
  3. Pick the output language and press Analyze.
  4. Go through the issues from Critical down; each points at a JSON path and quotes the value that shows the problem. Download as CSV, Markdown or JSON.

AI API Response → Potential Issues features

  • Spots inconsistent types, missing fields, wrong status semantics, leaked data, bad date formats, null handling, naming and pagination smells
  • Every issue carries a JSON path and quoted evidence
  • Optional expected contract to compare against
  • Health score plus "looks good" list

AI API Response → Potential Issues example

Response review

Input:

{"data":[{"id":1,"passwordHash":"…"},{"id":"2","createdAt":"02/01/2026"}]}

Output:

ISS-1 data-exposure (Critical) $.data[0].passwordHash
ISS-2 inconsistent-type (High) $.data[*].id — number vs string
ISS-3 bad-date-format (Medium) $.data[1].createdAt

Frequently asked questions about AI API Response → Potential Issues

Should I include headers?

Yes when you can — status and headers let the AI check status semantics, content types and caching in addition to the body.

What happens when I add the expected contract?

The response is compared with it, and every deviation — a missing field, a different type, an unexpected status — is reported as an issue.

How are severities assigned?

Critical is data exposure or data corruption, High is something that will break clients, Medium is a contract or quality problem, and Low is cosmetic.

Technical notes

Each issue has an ID, a kind (inconsistent type, missing field, wrong status, data exposure, bad date format, null handling, naming, duplicate, encoding, pagination, error format, performance or other), a severity, the JSON path, header or "status" it concerns, the evidence value and a suggested fix. The report adds a 0–100 health score and a "looks good" list; an issue that cannot be pointed at in the response is not reported.

References