Use when asked to review a local diff, a GitHub PR, or a whole codebase. Reports each finding with severity, likelihood, a worth-fixing verdict, and a high-level fix.
Use when asked to trim, clean up, shorten, or remove code comments. Two levels — "normal" (default) shortens verbose comments; "aggressive" deletes nearly every comment, keeping only trimmed unusual-workaround notes.
Use after generating or editing ASCII art, box diagrams, tables, or any monospace text art to ensure all lines, corners, and boxes are properly aligned.
Use when creating a new agent skill. Ensures the skill follows quality guidelines for SKILL.md files.
Use when updating GitHub PR title or body. Works around the gh pr edit GraphQL bug caused by GitHub's Projects Classic deprecation.
Use when the user wants to interact with Linear.app — reading or searching issues/tickets.
Use when writing or editing a system prompt for any LLM API or SDK (any code passing a `system=` / `system` role parameter, or a `.txt`/`.md` file holding such a prompt). Applies prompt-engineering and prompt-caching best practices.
Use when the user asks to explain or teach a technical concept. Replies in plain language with a simple diagram instead of a wall of jargon.