The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute a shell command using grep to scan local source files for sensitive strings like API keys and passwords.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it is designed to ingest and process untrusted external data (source code). Maliciously crafted comments or strings within the audited code could attempt to influence the agent's behavior during the audit process.
Ingestion points: Local source code files in the src/ directory.
Boundary markers: None provided to distinguish between instructions and data.
Capability inventory: Local file system read access via grep and review of code structures.
Sanitization: No sanitization or validation of the processed file content is described.
[SAFE]: The skill promotes security best practices, such as advising against direct shell concatenation in subprocess calls and recommending the use of dependency lockfiles.

sec-auditor