skills-curator

To evaluate third-party skills, you can run a security-first audit that scans your project signals and generates a structured ADOPT, PARTIAL, or SKIP verdict, complete with pros, cons, and an adoption plan.

The best way to customize an external skill is using a command-driven interface to decompose its SKILL.md into sections, generating a per-section action plan that rewrites examples into your stack's voice without inheriting the author's accent.

You can audit external skill folders for security conflicts by applying a registry-based verdict workflow that checks for duplicates and conflicts, requiring a YAML-frontmatter-defined entry point to process the evaluation.

Yes, you can migrate skills across agent platforms using deterministic command-driven interfaces for evaluation and export, which document the initial decision to avoid repeated decision-making and uncertainty during the migration.

Project-aware skill recommendation works by scanning your specific project signals to rank and suggest relevant skills, ensuring the discovery process is tailored to your environment rather than depending on general popularity metrics.

A key limitation is that the evaluation and curation process requires a YAML-frontmatter-defined entry point, meaning skills lacking this specific metadata structure cannot be properly audited or processed for adoption.