#flaky-tests

Turn failing pytest runs into structured repair reports: failure parsing, root-cause diagnosis, flaky detection, dry-run fix diffs, and an optional LLM layer. 把失败的 pytest 运行转化为结构化修复报告。

Claude Code QA agent with memory, plus a read-only MCP server (verdict-qa-mcp on PyPI): baseline → delta runs (NEW/REGRESSED), a fact harness so the model judges while the system measures, signed run history, flaky quarantine with expiry, and an exit-code release gate for CI. Ships six scored eval fixtures — misses published.

AI agents make CI pass. Sutura verifies the fix, filters flaky failures, rejects unsafe shortcuts, and opens an evidence-backed PR for human review.

Statistically-corrected, flakiness-aware regression detection for LLM agents, wired into a real CI-gate workflow and judged on traces and tool calls, not just final text. Runs every case many times and fails a pull request only when a drop is statistically meaningful, not when a flaky case happened to fail.

GitHub Action that groups failed tests by root cause and explains them on the PR