The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external source code, API responses, and dependency manifests, creating a surface for indirect prompt injection attacks.
Ingestion points: Targets specified in SKILL.md (e.g., src/ and https://staging.example.com), scanning scopes in references/compliance-agent-commands.md, and code fixtures within evals/security-testing.yaml.
Boundary markers: The instructions do not define explicit delimiters or instructions to ignore embedded commands within the analyzed data.
Capability inventory: The skill utilizes node -e for file system writes, docker run for dynamic scanning, and fetch and npm audit for network-based operations.
Sanitization: There is no evidence of sanitization or content validation performed on the external data before it is incorporated into the agent's context.
[DYNAMIC_EXECUTION]: SKILL.md contains a template for a node -e command block used to update the run-history.json file. This is categorized as simple script generation from templates for the purpose of local state management.
[EXTERNAL_DOWNLOADS]: The skill references well-known security resources, including the trufflehog GitHub Action and the owasp/zap2docker-stable Docker image. These are industry-standard tools from established organizations.
[COMMAND_EXECUTION]: The skill instructions incorporate the use of system commands such as npm audit, docker run, and project-specific CLI tools like aqe to perform its core security scanning and auditing functions.

security-testing