The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes scripts/verify.sh, a bash utility that uses grep to scan documentation files for a hardcoded list of 'weasel words' (e.g., 'simply', 'just'). The script is a read-only tool intended for local linting by the agent to ensure prose quality.
[EXTERNAL_DOWNLOADS]: The skill references established industry resources including the Google Developer Documentation Style Guide, the Diátaxis framework (diataxis.fr), and the Vale prose linter (vale.sh). It provides a sample GitHub Actions configuration that references the official errata-ai/vale-action@reviewdog action for automated CI linting.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and restructure technical documents provided by users.
Ingestion points: Technical documentation files (Markdown/reStructuredText) processed by the agent or passed as arguments to the scripts/verify.sh script.
Boundary markers: The skill enforces strict structural boundaries by requiring all documentation to fit exactly one of four Diátaxis modes (Tutorial, How-to, Reference, or Explanation).
Capability inventory: The skill's automated capability is limited to local file system read access for linting purposes. It does not perform network operations or execute arbitrary shell commands based on document content.
Sanitization: The prose verification script uses static regex patterns against a fixed, hardcoded banlist to validate content, preventing data from influencing the script's execution logic.

technical-writing