The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local script scripts/cognitive_load.py using a user-provided <file-path> as a command-line argument. This creates a risk of command injection if the input path is not sanitized before being passed to the shell.
[DATA_EXFILTRATION]: The skill is designed to read and analyze the content of local files and URLs provided by the user. This capability could be exploited to read and expose the contents of sensitive files (e.g., credentials, configuration files) if the agent is tricked into "analyzing" them as blog posts.
[PROMPT_INJECTION]: The skill processes untrusted data from external sources (local files and URLs) and incorporates this content into its internal analysis and scoring process. This constitutes an indirect prompt injection surface where malicious instructions embedded in a blog post could attempt to influence the agent's behavior or audit results.
Ingestion points: Local files read via argument-hint and URLs fetched using WebFetch (SKILL.md).
Boundary markers: None identified in the instructions to delimit untrusted content.
Capability inventory: File system read access, network access (WebFetch), and subprocess execution of scripts/cognitive_load.py (SKILL.md).
Sanitization: No explicit sanitization or validation of input content or file paths is described.

blog-analyze