The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes Python test scripts (scripts/test_repository_validator.py and scripts/test_template.py) that use subprocess.run to execute internal validation scripts and external CLI tools like skills-ref. This behavior is localized to the skill's own testing framework and is used to verify the integrity of skill templates and repository structure.
[INDIRECT_PROMPT_INJECTION]: As a maintenance and auditing tool, this skill is designed to process other skill packages, which could potentially contain malicious instructions.
Ingestion points: Reads skill directories, SKILL.md files, and bundled resources during audits.
Boundary markers: Relies on the normative specification for Agent Skills and encourages the use of standard templates.
Capability inventory: Uses subprocess.run for testing and recommends various linting/validation tools (ruff, shellcheck, pyright).
Sanitization: The documentation (references/audit-workflow.md) explicitly instructs the user to exercise changed scripts and templates in disposable copies and to verify technical claims against authoritative sources.
[EXTERNAL_DOWNLOADS]: The skill documentation (references/validation-tools.md) lists several well-known developer tools and their official repositories, including Ruff, Pyright, Hyperfine, and Actionlint. It also references VersionLens Redux, a VS Code extension authored by the skill's vendor (xsyetopz). These references are documented neutrally as part of the recommended validation workflow.

maintain-agent-skills