The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes findings and remediation data retrieved from external API sources, creating a potential surface for instructions hidden within analyzed data to influence the agent behavior.
Ingestion points: Data is fetched via API requests in assets/tool_implementation.py and converted to models for consumption.
Boundary markers: The implementation does not include explicit boundary delimiters or 'ignore' instructions for the data interpolated into the agent context.
Capability inventory: The skill is configured with access to high-capability tools such as Bash, Write, and Edit, as defined in the skill frontmatter.
Sanitization: Data from the API is mapped directly into LLM-facing models in assets/models.py without specific filtering or sanitization of text fields such as remediation recommendations.

prowler-mcp