The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill fetches an external dependency from the author's GitHub repository using git submodule add https://github.com/RoleModel/rolemodel-skills.git to provide the wrap-up skill functionality.
[COMMAND_EXECUTION]: During the verification step, the skill executes the newly installed Ruby hook script (assets/surface_conventions.rb) via a shell command to validate its parsing logic and ensure it does not raise errors.
[INDIRECT_PROMPT_INJECTION]: The skill installs a PreToolUse hook that automatically injects documentation content into the agent's prompt based on the files being edited. This creates a surface where malicious instructions placed in documentation files could influence agent behavior.
Ingestion points: The hook script reads data from docs/CONVENTIONS.md and individual convention files located in docs/conventions/*.md.
Boundary markers: The script uses a specific header ("📐 Convention docs that apply to...") to prefix the injected content, though it does not explicitly instruct the model to ignore potential instructions within that content.
Capability inventory: The skill and the agent environment it configures have access to file system tools (Read, Write, Edit, Glob, Grep) and shell execution (Bash).
Sanitization: The script parses the documentation for metadata and globs but does not perform content sanitization or safety filtering on the text before it is emitted as additionalContext.

scaffold-docs