The Agent Skills Directory

[COMMAND_EXECUTION]: The script eval/run_eval.py uses subprocess.run to execute local shell scripts (validate-skill.sh, prompting-playbook.sh, frontmatter-standard.sh) located in the repository's scripts/gates/ directory. These are used for validating that the scaffolded skill skeleton meets structural and quality requirements.
[COMMAND_EXECUTION]: The unit test file assets/test_scaffold_skill.py uses subprocess.run to execute the scaffolder script and the generated smoke tests. This is a standard practice for verifying the correctness of the tool's output.
[COMMAND_EXECUTION]: The SKILL.md file provides a shell command (find ~/.claude ~/.config ~/.agents -type d -name 'repo2skill') for the agent to locate the skill's installation directory. This command is used for discovery and does not involve network activity or pipe output to dangerous interpreters.
[SAFE]: The skill script assets/scaffold_skill.py is written using only Python standard libraries and operates offline. It includes strict regex validation (^[a-z0-9]([a-z0-9-]*[a-z0-9])?$) for the skill name to prevent directory traversal or command injection via malformed names.
[SAFE]: The logic for generating files uses string.Template for deterministic substitution of metadata (author, version, tags) into template blocks, which is a secure method for generating static content.

repo2skill