The Agent Skills Directory

[SAFE]: The skill operates as a static index and gateway for documentation. It does not execute code, perform network operations, or access sensitive system credentials.
[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to read external markdown files located in a relative directory (../mcp/*.md) based on user intent. While these files are intended documentation within the skill's environment, this pattern creates a surface for indirect prompt injection if the content of those files were to be maliciously modified by an external process.
Ingestion points: SKILL.md instructs the agent to read files such as ../mcp/INDEX.md and ../mcp/mcp-tool-design.md.
Boundary markers: No explicit boundary markers or 'ignore' instructions are provided for the content of the loaded files.
Capability inventory: The skill utilizes the agent's file reading capabilities.
Sanitization: No sanitization is performed on the content of the markdown files before it is loaded into the context.

discover-mcp