The Agent Skills Directory

[PERSISTENCE]: The skill installs a Node.js script as a PreToolUse hook in .claude/hooks/inject-code-style.js and modifies the project's .claude/settings.json to ensure this script executes every time the agent performs edit or write operations.
[DYNAMIC_EXECUTION]: The skill dynamically creates an executable script by interpolating runtime-detected file extensions into a predefined template (references/inject-code-style.template.js) and writing the result to the local filesystem.
[COMMAND_EXECUTION]: The instructions require the agent to execute shell commands to validate and test the newly generated script, including node --check and running the script with mocked JSON input to verify its output behavior.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: The skill reads project source code and linter configuration files across the repository to generate docs/PROJECT-CODE-STYLE.md (Step 1 and 2 in SKILL.md).
Boundary markers: The injected context generated by the hook (references/inject-code-style.template.js) lacks protective delimiters or instructions for the agent to ignore potentially malicious directions embedded in the codebase-derived style guide.
Capability inventory: The agent possesses extensive capabilities including file system access and command execution, which are active when the hook injects content during development tasks.
Sanitization: Content from the analyzed codebase is merged directly into the markdown documentation without validation or escaping, and subsequently injected into the agent's active context via the generated hook.

project-code-style-analyzer