The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes external source code and data for security scanning, creating an indirect prompt injection surface.
Ingestion points: SKILL.md (via the target parameter) and evals/security-testing.yaml (via code fixtures).
Boundary markers: No explicit boundary markers or instructions to ignore embedded commands are defined for the input data.
Capability inventory: Orchestrates security agents for SAST/DAST scanning, dependency audits, and coordinates multiple specialized security agents.
Sanitization: No explicit sanitization or input filtering logic is implemented in the skill body.
[CREDENTIALS_UNSAFE]: The evals/security-testing.yaml file contains hardcoded credentials and API keys (e.g., sk-1234567890abcdef, admin123). These are used as test fixtures for the tc005_hardcoded_credentials test case to verify the skill's detection capabilities and are not used for authentication in the skill logic itself.
[EXTERNAL_DOWNLOADS]: The skill configuration references official security tools and containers, such as the owasp/zap2docker-stable image and the trufflehog GitHub Action. These originate from well-known and trusted security organizations and are used for standard auditing purposes.

security-testing