llm-utils

Write and audit project documentation on the Diataxis model without tutorials: explanation, reference, and optional how-to, one topic per folder.

4 skills · 1 command

Patterns for AI-powered apps: retrieval evaluation, grounded generation with citations, SSE streaming UI, and the OpenAI API.

4 skills

Write and audit agent-facing documentation on a progressive-disclosure model: a short CLAUDE.md of critical rules that routes to detail loaded only when needed.

2 skills · 1 command

Staged delivery workflow with hard gates: exploration, PRD, spec, plan, tasks, implementation, feedback, review, finalize. Enforces ADR-per-decision and living specs.

1 skill · 1 command · hooks

Storybook component documentation and AG-UI protocol expertise for agent-connected UIs.

2 skills

The /mr command: audit a branch against project docs and ADRs, fix findings, commit without AI attribution, and open a GitHub PR from the repository template.

1 skill · 1 command

Reconstruct architectural decisions already embedded in an existing codebase and record them as ADRs.

1 skill

Keep a typed backend API and its frontend client in sync via OpenAPI codegen, with drift detection in CI.

1 skill

Delegate large-context, multimodal, and bulk-refactoring work to Gemini CLI, with token savings reported.

1 skill

Verify an implemented UI against its design spec: design-token compliance, responsive breakpoint behaviour, and interaction states.

1 skill

Run a retrospective on a coding session and suggest changes to the agent's environment: navigation pointers, automated checks, review rules, steering-file trimming.

1 command

Convert agent skills between Claude Code, OpenAI Codex, GitHub Copilot, and Gemini formats, and author skills that port cleanly.

1 skill

Turn a feature request into a minimal spec a developer can build from: the need, the scope boundary, and Gherkin acceptance criteria including the unhappy path.

1 skill