23 practical agent skills: TeamWorkflow, multi-model review, media ingestion, and writing. Original Claude Code packages plus a Codex desktop edition tuned for Astra. Source for clickai.dev.
This repository also configures its own agents. See what skills tells them →
Agent Claude Code
Use to find code and answer questions about it: where something is defined or used, which files match a pattern or naming convention, what a piece of code does. Name the breadth in the request: quick, medium, or very thorough. Read-only. Never for judgment, review, design, or changing files.
Agent Claude Code
Use only for a step whose request names its pass/fail check: bring a branch up to date with main, run named verify commands and report each exit code, apply a specified edit or rename, or fix a failure whose cause is already named. Never for design, building a feature, review, or any step that needs a decision.
Agent Claude Code
Use for a question answered by reading: files, issue and PR state, CI logs, command output, docs. Returns the facts and where it found them. Never for judgment, review, design, or changing files; that work stays on the session's model.