The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The scripts/security_scan.py script processes files within a user-provided project directory. This represents a potential surface for indirect prompt injection if malicious files are crafted to influence the agent's interpretation of the scan results.
Ingestion points: Files within the directory specified by the project_path argument in scripts/security_scan.py are read and analyzed.
Boundary markers: No specific delimiters or instructions are used to prevent the agent from misinterpreting embedded instructions found within scanned files.
Capability inventory: The script uses subprocess.run to call npm audit and performs standard file reading operations.
Sanitization: The script uses regular expressions for pattern matching and json.loads() for parsing tool output, but does not sanitize the raw content of the files it scans.
[COMMAND_EXECUTION]: The script scripts/security_scan.py uses subprocess.run to execute the npm audit command.
The execution is used to perform dependency auditing as part of the skill's stated purpose.
The command npm audit --json is executed within the context of the user-supplied project_path using the standard library subprocess module.

vulnerability-scanner