dual-axis-skill-reviewer

Automate skill quality review by running run_dual_axis_review.py with the --project-root argument, which executes deterministic checks and LLM evaluation to produce a weighted score for one or more project repositories.

Dual-axis evaluation combines deterministic checks on repository artifacts and SKILL.md with structured LLM JSON review, merging both axes into a single weighted score with actionable improvement items for skills scoring below 90.

Cross-project skill reviews are supported using the --project-root command line argument, allowing you to target and evaluate skills located in another local repository directory.

Yes, pytest and uv are required dependencies for executing test health checks and running the reviewer script, which requires Python 3.9 or higher to parse YAML frontmatter and inventory repository artifacts.

Automated LLM evaluation requires strict JSON merging for weighted score computation and depends on valid YAML frontmatter parsing, meaning malformed metadata or non-compliant LLM JSON output can disrupt the scoring process.