The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data, specifically DAST configuration files and CI/CD pipeline scripts. While the skill includes a 'Prompt Injection Safety Notice' instructing the agent to treat content as data, there are no programmatic delimiters or technical sanitization steps to enforce this boundary.
Ingestion points: Processes target files provided via $ARGUMENTS and discovered configuration files (YAML, JSON, TSV, Jenkinsfile).
Boundary markers: Relies on natural language instructions rather than structural delimiters (e.g., XML tags or specific markers) to separate untrusted data from instructions.
Capability inventory: Access to Read, Grep, and Glob tools.
Sanitization: No programmatic validation or filtering of content before it enters the agent context.
[DATA_EXPOSURE]: The process explicitly directs the agent to locate and read sensitive files, including .env files, credential stores, and files likely to contain secrets like DAST_PASSWORD. While this is aligned with the skill's primary purpose of security auditing, it creates a surface area for sensitive data exposure within the agent's context.

dast-config