The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources to provide context for its decision-making agents.
Ingestion points: The skill instructions direct the agent to "pull any context you have (files, prior messages, project notes)" when a trigger phrase is detected (SKILL.md).
Boundary markers: There are no specific instructions or delimiters provided to ensure the agent ignores malicious instructions that might be embedded within the project files or notes.
Capability inventory: The skill uses the agent/Task tool to spawn five independent subagents that execute logic based on the ingested context (SKILL.md).
Sanitization: No explicit sanitization or filtering of external content is performed before it is passed to the subagents, although the multi-stage peer-review process may provide some inherent resistance to outlier behaviors.

llm-council