The Agent Skills Directory

[SAFE]: The primary purpose of the skill is to guide an agent in creating structured onboarding documentation, which is a benign and productive use case.
[COMMAND_EXECUTION]: The Workflow section in SKILL.md includes instructions to run basic discovery commands such as find, cat package.json, and cat .env.example. These are standard, read-only reconnaissance commands used to understand the project structure and are necessary for the skill's stated purpose.
[PROMPT_INJECTION]: The skill processes untrusted project data (e.g., file names and configuration templates) to generate README content, creating an indirect prompt injection surface.
Ingestion points: The Workflow section reads local directory structures and the contents of package.json and .env.example (SKILL.md).
Boundary markers: None present in the instructions.
Capability inventory: The skill uses Read, Glob, Grep, Edit, Write, and Bash tools.
Sanitization: No specific sanitization of the project data is mentioned.

readme-for-developers