v1
latestOpenAPI 3.0.22026-07-23125563.8 KBGet test failure context
Retrieve structured test failure context for AI debugging. One of three query shapes is used:
- Run-level: run_id only (omit instance_id and test_id). Paginate failed tests with limit and page.
- Instance-level: run_id and instance_id (omit test_id). Paginate with limit and page.
- Test-level: instance_id and test_id. The run is resolved server-side from the instance; optional run_id is ignored for lookup. Optional attempt selects a specific attempt; omit for the latest.
Responses can be large (for example, tests with many steps). There is a 1MB response limit. Include Accept-Encoding: gzip in your request when possible. You can also use detail=compact or detail=summary, or (for format=md) max_length, to reduce response size.
Query parameters
Run identifier. Required for run-level (run_id only) and instance-level (run_id + instance_id, no test_id). For test-level (instance_id + test_id), ignored; the server resolves the run from the instance.
Instance identifier. Required for instance-level and test-level. Omit for run-level (use run_id only).
Test identifier. When set, selects test-level detail and requires instance_id.
Attempt number (0-indexed); defaults to latest
Response format. Falls back to Accept header when absent.
Controls output verbosity to limit token count. 'default' returns all available data. 'compact' reduces output size by omitting the full steps array and limiting assets to 3 per category while preserving failure context and analysis. 'summary' minimizes output by including only error info and metadata.
Maximum number of failed tests per page (run-level and instance-level only)
Page number for failed tests pagination, 0-indexed (run-level and instance-level only)
Truncate markdown response to this character limit (only applies when format=md). Useful for fitting context into AI prompts.
Response
Context data