The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local shell commands via the 'claude' CLI and 'lsof' utility.
Evidence: Found in scripts/run_eval.py, scripts/improve_description.py, and eval-viewer/generate_review.py using subprocess.run and subprocess.Popen.
Purpose: These commands are used to test skill triggering, optimize descriptions using the model, and manage ports for the local evaluation viewer server.
[EXTERNAL_DOWNLOADS]: The evaluation viewer references an external library from a well-known CDN.
Evidence: eval-viewer/viewer.html includes a script tag fetching the SheetJS library from cdn.sheetjs.com.
Purpose: This library is used to render Excel (.xlsx) files within the local evaluation report for user review.
[COMMAND_EXECUTION]: The skill modifies file permissions for newly created scripts.
Evidence: scripts/init_skill.py uses chmod(0o755) on generated example scripts.
Purpose: This is a standard operation to ensure that initialized scripts are executable by the user.

skill-creator