The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill presents an attack surface where malicious instructions embedded in a project's source code or documentation could influence the agent's behavior during the doc-writing or refactoring process.
Ingestion points: The agent is instructed to read local source code files (CLI definitions, MCP signatures, API route handlers, config schemas) and existing documentation files to verify facts and perform refactors (SKILL.md, Step 4 and Step 5).
Boundary markers: Absent. The instructions do not define delimiters or provide specific warnings to the agent to disregard potential instructions embedded within the ingested source code comments or docstrings.
Capability inventory: The skill utilizes file system read/write operations and can execute shell commands (specifically running --help on project CLI tools as directed in Step 4). It also utilizes MCP tools for diagram generation.
Sanitization: Absent. There is no mention of sanitizing or escaping content retrieved from the source code before it is incorporated into the generated documentation or used in decision-making.

doc-write