The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill reads untrusted data from the repository's codebase and documentation to inform its design challenges, which creates a vulnerability where malicious content in those files could influence the agent's logic or file-writing actions.
Ingestion points: SKILL.md instructs the agent to 'Read the plan, linked docs, code, or local files needed to understand the topic' and to 'explore the codebase' to answer questions.
Boundary markers: No specific delimiters or instructions are provided to the agent to distinguish between its primary instructions and the content of the files it is reading.
Capability inventory: The skill is authorized to write and modify files within the repository, specifically CONTEXT.md, CONTEXT-MAP.md, and Architecture Decision Records (ADRs).
Sanitization: The skill does not include instructions for sanitizing, validating, or filtering the content read from the repository before it is processed or written into new documents.

grill-me