The Agent Skills Directory

[DATA_EXFILTRATION]: The pre-scan.sh script specifically searches for sensitive file paths such as .env, *.pem, *.key, and credentials.json. While intended to identify files accidentally committed to version control, this functionality allows for the discovery of credentials and secrets within the environment.
[PROMPT_INJECTION]: The skill processes untrusted repository content to identify cleanup candidates, which presents an indirect prompt injection surface.
Ingestion points: Repository code files are read by rg and Python's ast module in pre-scan.sh for analysis.
Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the skill's logic.
Capability inventory: The skill is granted Read, Write, Bash, Glob, and Grep tools, allowing for significant file system modification.
Sanitization: No sanitization or validation is performed on the repository data before it is processed by the agent.

trash-truck