The Agent Skills Directory

[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it ingests and processes source code from external projects to inform its analysis logic.\n- Ingestion points: Codebase files are read and analyzed during the Phase 1 reconnaissance and Phase 2 verification steps.\n- Boundary markers: The skill uses structured instruction sets for subagents, separating task descriptions from the data context using clear section headers.\n- Capability inventory: The tool can read and write files within the project's 'sast/' directory and spawn multiple subagents for parallel processing.\n- Sanitization: The instructions do not specify any sanitization, validation, or escaping of the code snippets retrieved from the project before they are passed to subagents.

sast-graphql