The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted code repositories which serves as a vector for indirect prompt injection. Maliciously crafted source files could attempt to influence the agent's documentation output or behavior. * Ingestion points: The skill uses glob and read tools to ingest all files within a target repository in the SKILL.md workspace exploration phases. * Boundary markers: No specific boundary markers or instructions to ignore embedded commands are used when processing file content. * Capability inventory: The agent has the capability to read any file in the repository and write a new Markdown document to the filesystem. * Sanitization: No sanitization of the ingested code content is performed before it is used to generate the report.
[DATA_EXPOSURE_AND_EXFILTRATION]: The instructions in Phase 2 of SKILL.md direct the agent to specifically identify and read sensitive configuration files such as .env, settings, and config directories. While intended to document project architecture, this behavior risks extracting and exposing real secrets or credentials if the agent includes actual values in the final generated documentation file instead of just structural descriptions.

project-analyzer