Conduct a retrospective on a coding session and turn it into ranked proposals for the agent's environment: navigation pointers, automated checks, reviewer rules, leaner steering files, cheaper tool calls. Invoke as /retro.
Produce publication-quality technical figures: a self-contained HTML preview plus a single-file SVG. Covers architecture topology, data flow, layers/nesting, loop/flywheel, sequence, request lifecycle, state machine, swimlane, benchmark bars, trend/percentile…
Design and review deep modules: choose the interface, place the seam, and shape what sits behind it. Load it when code changes shape rather than behaviour: moving logic out of `main` or a script into a module, splitting one function or class into several,…
Generate a project-local verification skill that drives your app the way a user does — any language, framework, or platform. Use for /create-verification-skill, "make a control skill for this repo", or when a project has no scripted way to prove…
Skill authoring and agent-facing docs: make a SKILL.md, an AGENTS.md / CLAUDE.md, or any document an agent reaches by pointer fire reliably and stay short. Use when creating or editing a skill, writing or pruning agent instructions, choosing model- vs…
Test whether each component, layer, or mechanism earns its place by removing or replacing it and comparing the result against a baseline. Use when designing or choosing between system architectures, judging whether an existing system is over-engineered,…
Remove AI-generated code slop from a branch diff. Load it before you commit, open a PR, or report a code change as complete, even when the user does not mention slop or style. It reads the diff against main and strips what a model added without being asked:…
Cut AI tells from any writing: AI vocabulary, chatbot phrasing, rule-of-three, em dashes, boldface, hedged generalities. Must always apply, so load it before you write or edit prose the user will read, even when they do not ask for a rewrite or mention style:…