The Agent Skills Directory

[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it is designed to ingest and process information from other files in the repository to explain system behavior. This could allow malicious instructions embedded in those documentation files to influence the agent.
Ingestion points: In Phase 2, the skill uses the Read tool to fetch content from skills/{skill-name}/SKILL.md and agents/{agent-name}.md.
Boundary markers: The instructions do not define clear delimiters or include 'ignore instructions' headers when the agent processes the retrieved file content.
Capability inventory: The skill has access to the Read, Grep, and Glob tools.
Sanitization: There is no specified logic for sanitizing or filtering the content of documentation files before the agent analyzes them to provide an explanation.

workflow-help