The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses platform-specific shell commands to manage documentation files. It executes rm and ln (Unix) or Remove-Item and New-Item (PowerShell) to delete existing files and create symbolic links for CLAUDE.md pointing to AGENTS.md in the local directory.
[PROMPT_INJECTION]: The skill reads data from potentially untrusted project files (such as README.md, .cursorrules, and Copilot instructions) to populate the AGENTS.md file. This creates a surface for indirect prompt injection if those files contain malicious instructions.
Ingestion points: README.md, .cursor/rules/, .cursorrules, and .github/copilot-instructions.md (processed in SKILL.md).
Boundary markers: Not utilized; the skill does not wrap ingested content in safety delimiters.
Capability inventory: Local file system manipulation (delete/symlink).
Sanitization: None; instructions direct the agent to include "important parts" of the found documentation directly.

init