gh-bench
gh CLI extension that extracts CI metrics and surfaces test failures from GitHub Actions logs, with framework-aware parsers and statistical aggregation.
stack
Goabout
i built gh-bench to make it easier to get metrics and failures out of GitHub Actions logs whilst staying in my terminal. gh bench stats pulls a regex-captured value from logs across many runs and aggregates it (median, mean, p95, min, max); gh bench failures fetches failed runs, identifies the failing steps, and extracts structured errors using framework-aware parsers for .NET, Go, and Vitest/Jest, falling back to cleaned error output when no framework is detected. gh bench grep searches run logs by keyword or regex across many runs, with context lines and failed-only scoping. by default it scopes to runs from branches with open pull requests, handles re-run attempts, and emits table, JSON, or CSV.
features
- gh bench stats: extract a regex-captured metric across runs and aggregate with median, mean, p95, min, and max
- gh bench failures: framework-aware failure extraction for .NET (xUnit/NUnit/MSTest), Go, and Vitest/Jest
- gh bench grep: search CI run logs by keyword or regex, with context lines and failed-only scoping
- open-PR scoping by default, with re-run attempt handling and per-failure CI timestamps
- fallback parser that strips Actions boilerplate and prioritises ##[error] messages
- project-level defaults via .gh-bench.yml, plus table, JSON, and CSV output