The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute local project tests and software commands (e.g., br list, bv --robot-insights) to verify the behavioral reality of the implementation and track project progress.
[DATA_EXPOSURE]: The skill reads internal project files, including documentation and source code, to perform its audit functions. Findings are stored locally within the project's .agents directory.
[PROMPT_INJECTION]: The skill identifies an indirect prompt injection surface as it processes potentially untrusted project documentation (README.md, PRODUCT.md).
Ingestion points: README.md, PRODUCT.md, and source code files are read from the repository.
Boundary markers: The skill does not explicitly define delimiters for untrusted content, though it requires the agent to distill it into falsifiable promises.
Capability inventory: The skill allows subprocess calls for running tests and writing report files to the local filesystem.
Sanitization: No explicit sanitization or filtering of external content is described before processing.

reality-check