The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions direct the agent to execute shell commands to verify technical consistency and build stability. Specifically, it uses grep to check for framework usage across chapters (Q10) and uv run mkdocs build --clean to ensure the documentation website generates without errors (Q14).
[INDIRECT_PROMPT_INJECTION]: The skill's core functionality involves ingesting and analyzing untrusted Markdown files provided as tutorial content. This creates a surface where malicious instructions embedded in the tutorials could influence the agent's review logic or scoring.
Ingestion points: The agent processes input .md files as the primary data source for its quality gates (SKILL.md).
Boundary markers: There are no explicit delimiters or instructions telling the agent to treat the tutorial content as data only or to ignore any embedded commands.
Capability inventory: The skill possesses the ability to execute system commands like grep, mkdocs, and uv (references/quality-gates.md).
Sanitization: No sanitization or safety filtering of the input Markdown files is specified before the agent performs its analysis.

tutorial-writer-review