The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute Node.js scripts and npm commands native to the repository, such as scripts/install-plan.js, scripts/install-apply.js, and npm run harness:audit. These are standard tools for managing the Everything Claude Code environment.
[INDIRECT_PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it reads and searches files within the repository that could contain untrusted instructions.
Ingestion points: Files located in skills/, commands/, agents/, and docs/ directories.
Boundary markers: The skill does not explicitly define delimiters to distinguish between instructional content and raw data.
Capability inventory: Execution of local Node.js scripts, npm scripts, and filesystem search utilities.
Sanitization: No specific content sanitization or validation mechanisms are implemented.

ecc-guide