GitHub Actions Log Grabber

Use the gh-logs-grab CLI tool to download CI logs from GitHub Actions and analyze pytest failures.

When to Use

• •User shares a GitHub Actions run URL (e.g., https://github.com/owner/repo/actions/runs/123456)
• •User asks about CI failures, test failures, or flaky tests
• •User wants to compare errors across multiple CI runs
• •User asks about job timing/performance in CI

Commands

Download Logs

gh-logs-grab download <RUN_URL> [OPTIONS]

Downloads logs from a GitHub Actions run. By default, only failed jobs are downloaded.

Options:

• •-o, --output <PATH>: Output directory (default: ./logs)
• •-a, --all: Include successful jobs too
• •-t, --token <TOKEN>: GitHub token (auto-detects from GITHUB_TOKEN env or gh auth token)

Output structure:

logs/pr-{number}/{run-id}/
├── metadata.json
├── findings.json          # Auto-generated pytest analysis
├── job-name-failure.log
└── ...

Analyze Single Run

gh-logs-grab analyze <RUN_DIR>

Extracts pytest errors from logs. Creates findings.json with deduplicated errors.

Note: download runs this automatically.

Timeline Analysis

gh-logs-grab timeline <PR_DIR>

Compares errors across multiple runs in a PR directory. Creates analysis.json with:

• •Regressed: New errors introduced
• •Persistent: Present in all runs
• •Intermittent: Flaky tests
• •Fixed: Errors that disappeared

Job Timings

gh-logs-grab timings <PR_DIR>

Analyzes job duration across runs. Creates timings.json and shows top 5 slowest jobs.

Typical Workflow

1. •User shares run URL → gh-logs-grab download <url>
2. •Review findings.json for error summary
3. •If multiple runs exist → gh-logs-grab timeline logs/pr-XXX to identify regressions vs flaky tests
4. •For performance issues → gh-logs-grab timings logs/pr-XXX

Notes

• •Logs are stored in ./logs/ relative to current directory
• •Tool is idempotent - safe to re-run without re-downloading
• •GitHub token is auto-detected from environment or gh CLI