Debugging & Diagnosis with AI
Ora can analyze test failures, offer fix suggestions, and explain API errors — reducing the time spent diagnosing issues. Failures are diagnosed by the Execution specialist; documentation questions are answered directly by the Orchestrator using search_documentation.
Analyzing Test Failures
After running a test that fails:
"Why did this test fail?"
"Analyze the last test run and suggest fixes"
The Execution specialist reads the test report, identifies the failing step, examines the request/response that caused the failure, and suggests specific corrections. It never modifies an assertion just to accept an error response — it diagnoses the root cause instead.
Diagnosing HTTP Errors
"Why am I getting a 401 on the /orders endpoint?"
"What does this 500 error with traceId abc123 mean?"
Ora checks:
- Authentication configuration (missing headers, expired tokens)
- Request format (wrong Content-Type, malformed body)
- Environment settings (wrong base URL, port)
- Server-side issues from error response bodies
Fixing Assertions
"This assertion fails because the response structure changed — update it"
"Relax this assertion to only check status code, not the body"
Ora updates the assertion step to match the current response structure.
Searching Documentation
For how-to questions about Aethrix Studio features, Ora searches the product documentation itself (via search_documentation) and answers in natural language:
"How do I set up Kafka authentication?"
"Show me how to use data transfer between steps"
"What's the difference between environment variables and project variables?"
General API concepts, protocols, or external services are answered from Ora's own knowledge.
Understanding Test Reports
"Summarize the test report — which suites had the most failures?"
"Are there any patterns in the failed tests?"
For load tests, Ora reports percentiles (p50/p90/p99), requests per second, and error rate; for data-driven tests it reports per-row status.
Tips
- Include the specific error message or status code in your prompt.
- Reference the test case name or step when asking about a specific failure.
- For complex issues, provide the sequence of steps leading to the error.