The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill scans untrusted project files like API specs, Dockerfiles, and infrastructure manifests to build data flow diagrams, which presents a surface for indirect instructions to influence the agent.\n- Ingestion points: Scanning route definitions, API gateway configs, and deployment manifests as defined in the workflow steps.\n- Boundary markers: The instructions lack explicit delimiters or warnings to the agent to disregard any instructions found within the analyzed files.\n- Capability inventory: The skill is designed for analysis and documentation and does not include capabilities for command execution or network access.\n- Sanitization: The workflow does not specify any sanitization or filtering of the information extracted from the external files.

pasta-scope