The Agent Skills Directory

[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it processes arbitrary code content provided through the ${file} and ${selection} variables.
Ingestion points: External code content is interpolated into prompts/generate-docs.prompt.md for analysis and documentation generation.
Boundary markers: Absent. The prompt does not utilize delimiters (e.g., XML tags or backticks) to isolate the code content or include instructions for the agent to ignore embedded commands.
Capability inventory: The agent has access to search, edit, and read tools, which could be abused if malicious instructions are followed.
Sanitization: Absent. There is no evidence of input validation or escaping of the code being processed.

Documentation Skill