What problem does it solve?
Investigating why CI is red on a Materialize PR requires digging through Buildkite builds, job logs, annotations, and artifacts, which is slow and error-prone when done manually. This Skill provides a structured triage workflow that finds the failing build, reads pre-extracted error annotations, fetches logs only when needed, and groups failures by root cause.
Core Features & Use Cases
- Build and job discovery: Locate failing builds from a PR number, GitHub URL, or Buildkite URL, and list genuinely failed jobs with correct job-state semantics (failed/timed_out vs. broken or waiting_failed).
- Annotation-first triage: Read pre-extracted error annotations with known-issue links (Linear keys like CPU-170) before touching raw logs, saving significant investigation time.
- Root-cause categorization: Apply Materialize-specific patterns for clippy errors, testdrive cascades, SLT failures, timeouts, feature benchmark regressions, cargo-fuzz crashes, and Miri failures.
- Known vs. new failure analysis: Use bin/ci-failures history, git log against main, and open PRs to distinguish pre-existing flakes from regressions introduced by the change.
- Use Case: A developer sees failing checks on PR #12345 and asks why CI is red; the Skill identifies the failed jobs, finds the annotation linking to a known flaky test, and reports that only one clippy error is a genuine new issue to fix.
Quick Start
Ask the assistant to investigate why CI is failing on PR 12345 or on a pasted Buildkite build URL.
