The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data which could be used as a vector for indirect prompt injection attacks.- Ingestion points: The skill reads the canonical QA Runbook, Ledger, Apex Scaffolding output, and repository diffs (SKILL.md).- Boundary markers: There are no explicit markers or instructions provided to the agent to distinguish between data content and instructions.- Capability inventory: The skill invokes tools such as ce-dogfood and agent-browser to perform actions in the environment (SKILL.md).- Sanitization: No sanitization or validation mechanisms are described for the external content before it is processed by the agent.

sociuu-aq