The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted markdown and text files from the documentation directory to extract requirements, which are then used to guide the agent's code verification process. This pattern creates a surface for indirect prompt injection where an attacker could embed malicious instructions in documentation.
Ingestion points: The skill reads all .md and .txt files in the ../docs/ folder.
Boundary markers: No specific delimiters or instructions are provided to separate analyzed content from the agent's operational instructions.
Capability inventory: The skill possesses capabilities to read the code base, generate markdown reports, and rename or move files to an archive directory.
Sanitization: There is no explicit requirement for the agent to sanitize or filter out instructions found within the requirements documents before performing verification tasks.

docs-analysis