The Agent Skills Directory

[SAFE]: The skill is a standard development utility for linting markdown files. It uses a local Python script to perform structural checks (e.g., frontmatter validity, line counts, and description quality).
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted markdown files, which introduces a theoretical surface for indirect prompt injection. However, this risk is mitigated by the following factors:
Ingestion Points: Files are read using Path.read_text() in scripts/lint_skills.py for static analysis.
Boundary Markers: Instructions in SKILL.md direct the agent to present linter output "as-is" without reformatting or interpretation, reducing the likelihood of the agent following instructions embedded in the linted content.
Capability Inventory: The skill is restricted to Read and Bash(python *) for the linter script. It does not have Write access or network capabilities, preventing data modification or exfiltration.
Sanitization: The script utilizes yaml.safe_load() for frontmatter parsing, protecting against YAML-based code execution vulnerabilities.
[DYNAMIC_EXECUTION]: While the skill executes a Python script, the script path is relative to the internal skill directory ${CLAUDE_SKILL_DIR}, and the script itself does not employ eval(), exec(), or other unsafe dynamic execution patterns. It uses standard regex-based validation logic.
[COMMAND_EXECUTION]: The skill uses the Bash tool to run its own bundled Python script. This is a primary function of the skill and is constrained to the specific linting script provided in the package.

ds-lint