The Agent Skills Directory

[SAFE]: No malicious behavior or security issues were identified. The skill implements functional security scanning tools (security_scanner.py, dependency_auditor.py, code_security_scanner.py) designed for authorized auditing and vulnerability assessment.
[COMMAND_EXECUTION]: The skill executes established security tools via the shell. Specifically, code_security_scanner.py runs the bandit static analyzer, and dependency_auditor.py wraps npm audit and pip-audit to perform vulnerability checks. These operations are essential to the skill's purpose.
[EXTERNAL_DOWNLOADS]: The setup_pentest_env.sh script installs project dependencies via pip install -r requirements.txt. Additionally, dependency_auditor.py includes logic to automatically install the well-known pip-audit package if it is not found on the system during execution.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data (HTTP responses from target URLs and local source code files) during scans. This represents a potential attack surface for indirect prompt injection if the scanned content contains instructions intended to mislead the agent's analysis.
Ingestion points: Reads HTTP headers/bodies from user-supplied URLs and reads text from local project files for static analysis.
Boundary markers: None implemented; scanner findings are returned to the agent context as structured Markdown or JSON.
Capability inventory: The skill has access to bash command execution and file read/write operations.
Sanitization: Content read from files or URLs is used directly in reporting without explicit sanitization or filtering for LLM instructions.

performing-penetration-testing