The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/scan.py script invokes git ls-files using the subprocess module. This is used for efficient file discovery but represents an execution of an external system binary.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it instructs the agent to analyze and summarize docstrings and comments extracted from the scanned codebase.
Ingestion points: Source code files are read by scripts/scan.py to populate digest.md for the agent.
Boundary markers: There are no specialized delimiters to isolate untrusted code comments from agent instructions in the digest.md briefing.
Capability inventory: The agent can execute local scripts and perform file system operations within the repository scope.
Sanitization: Extracted text is formatted for length but not sanitized for instructional triggers.

codegraph