conversational-api-debugger

4 Powerful MCP Tools

1. load_openapi

Load and parse OpenAPI 3.x specifications.

{
  "filePath": "/path/to/openapi.yaml",
  "name": "my-api"
}

Returns:

• API title, version, description
• Base URL and servers
• Complete endpoint list
• Endpoint count and structure

2. ingest_logs

Import HTTP request/response logs for analysis.

{
  "filePath": "/path/to/requests.har",
  "format": "har"
}

Or provide logs directly:

{
  "logs": [
    {
      "timestamp": "2025-10-10T12:00:00Z",
      "method": "POST",
      "url": "https://api.example.com/users",
      "statusCode": 400,
      "requestBody": { "name": "John" },
      "responseBody": { "error": "Missing required field: email" }
    }
  ]
}

Returns:

• Total requests ingested
• Success/failure breakdown
• Status code distribution
• Method distribution
• Top errors (first 10)

3. explain_failure

Analyze why an API call failed.

{
  "logIndex": 0,
  "specName": "my-api"
}

Returns:

• Severity: critical | high | medium | low
• Possible Causes: List of likely root causes
• Suggested Fixes: Actionable remediation steps
• Matching Endpoint: Comparison with OpenAPI spec
• Details: Request/response for inspection

4. make_repro

Generate cURL command to reproduce API call.

{
  "logIndex": 0,
  "includeHeaders": true,
  "pretty": true
}

Returns:

• cURL Command: Ready to copy-paste
• HTTPie Alternative: Shorter syntax
• JavaScript fetch: For automated tests
• Metadata: Method, URL, headers count