The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses mkdir -p to recursively create a predefined directory tree for documentation (e.g., docs/specs, docs/standards). This is a standard and expected operation for a project initialization tool.
[INDIRECT_PROMPT_INJECTION]: The skill populates a CLAUDE.md file, which is a known mechanism for influencing AI agent behavior within a workspace. While this creates an attack surface, the provided template content (templates/standards-claude.md) is limited to architectural standards and documentation naming conventions.
Ingestion points: Reads from the local templates/standards-claude.md file within the skill's package.
Boundary markers: None present in the output file, though the content is descriptive documentation rather than executable instructions.
Capability inventory: Limited to mkdir (directory creation) and file write operations on the local file system.
Sanitization: The skill performs an idempotent check, ensuring it does not overwrite existing user modifications to the CLAUDE.md file.

init-docs