What problem does it solve?
Post-release tasks like running evaluation baselines, updating dashboards, and extracting metrics for changelogs are critical but often manual, time-consuming, and prone to errors. This Skill automates these essential steps, ensuring accurate and up-to-date post-release documentation.
Core Features & Use Cases
- Automated Eval Baselines: Run comprehensive evaluation baselines with all production models for both AILANG and Python, providing robust performance data.
- Dashboard Updates: Automatically update the website benchmark dashboard, preserving historical data and clearing caches, ensuring the public-facing metrics are current.
- Changelog Metric Extraction: Extract and format key benchmark metrics for easy inclusion in
CHANGELOG.md, including automatic comparison with previous versions, eliminating manual data entry.
- Use Case: After a successful release, when a user says "run post-release tasks for v0.3.14," this Skill will execute the full eval baseline, update the website dashboard, and provide a ready-to-paste changelog entry, ensuring all post-release documentation is accurate and up-to-date.
Quick Start
Run evaluation baseline for a release version (all models)
.claude/skills/post-release/scripts/run_eval_baseline.sh 0.3.14 --full
Update website benchmark dashboard with new release data
.claude/skills/post-release/scripts/update_dashboard.sh 0.3.14
Extract benchmark metrics for CHANGELOG.md
.claude/skills/post-release/scripts/extract_changelog_metrics.sh