What problem does it solve?
This Skill autonomously tests Lark scenarios, analyzes failures, and iteratively repairs issues by tiered guidance.
Core Features & Use Cases
- Phase 1: Execute tests and collect outputs from the Lark testing suite.
- Phase 2: Analyze failures and identify root causes across scenarios.
- Phase 3: Classify failures into categories (test_drift, prompt_issue, tool_bug, gateway_logic, context_issue, llm_quality, architecture).
- Phase 4: Prioritize fixes by tier (Tier 1-4) with automated or semi-automated remediation workflows.
- Phase 5: Auto-iterate up to 3 rounds, re-running the test suite after each change.
- Phase 6: Generate a structured test report with summary, failure details, and repair history.
Quick Start
Run the Lark self-test workflow to execute the scenario tests, analyze failures, and generate a report.