The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to read various files within the target repository (such as package.json, source code, and configuration files) to verify claims for the README. This creates a surface where malicious content in those files could influence the agent's behavior.
Ingestion points: Reads repository files like package.json, Makefile, and LICENSE (specified in SKILL.md).
Boundary markers: No explicit delimiters or instructions are provided to the agent to ignore potentially malicious embedded content within the source files.
Capability inventory: The agent can read files, write markdown, and execute a provided Node.js script.
Sanitization: The skill does not define specific sanitization or escaping rules for data extracted from project files before interpolation into the prompt.
[COMMAND_EXECUTION]: The skill includes a Node.js utility script (scripts/check_readme.mjs) that the agent is expected to run to validate README files. The script performs local file system operations to check for dead links and formatting issues.
Evidence: SKILL.md instructs the agent to run node <本 skill 目录>/scripts/check_readme.mjs.
[EXTERNAL_DOWNLOADS]: Documentation templates provided for CLI tools and self-hosted apps contain example commands for downloading and piping scripts to a shell.
Evidence: references/cli-devtools.md includes curl -LsSf https://your-domain.sh/install.sh | sh as a template placeholder.

github-readme