The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/security_scan.py script invokes external security tools (bandit, pip-audit, semgrep, detect-secrets) using subprocess.run. The script follows security best practices by passing arguments as a list rather than a single string, which prevents shell injection vulnerabilities.
[INDIRECT_PROMPT_INJECTION]: The skill acts as a security scanner, which involves ingesting and analyzing untrusted code from a target project. While this creates an attack surface for indirect prompt injection (e.g., if a scanner's output is processed by an LLM), the implementation uses standard CLI tools and safe execution patterns to minimize risk.
Ingestion points: The scripts/security_scan.py script accepts a directory path and executes tools that read all files within that path.
Boundary markers: None observed in the script, though the primary interaction is via standard tool output.
Capability inventory: The skill has the capability to execute specific binary tools found on the system PATH and write a JSON report to the filesystem.
Sanitization: Arguments are passed via list-based execution in subprocess.run, preventing command injection from directory names or file paths.

auditing-python-security