amhuppert/ai-resources

Create a research or review agent for the /design workflow.

Assess / rate the default project context (CLAUDE.md, onboarding). Scores how well it conveys the project to AI across six dimensions, with improvement recommendations.

Automate browser interaction with Playwright: open pages, take screenshots, verify UI, test flows, debug pages, inspect elements.

Write a reviewer's guide to a set of code changes: purpose, the fundamental decisions, a file-by-file review path, and the diff noise to skip.

Review the Claude Code changelog and summarize recent changes and their significance.

Characterize what a codebase is built for: infer its design envelope from the mechanisms in the code, compare it with the operating envelope it actually faces, and report where the code is overbuilt or underbuilt.

Create a 1-2 page cheat sheet / quick reference for a tool, language, or framework.

Review a codebase for adherence to code standards. Uses specialized agents with progressive disclosure to determine which standards apply.

Compare two reports written by different AI agents from the same prompt — score both across accuracy, prompt adherence, organization, readability, and usefulness, then produce a polished self-contained HTML comparison report that declares a winner. Use whenever the user wants to compare, judge, grade, rank, or pick between two AI-generated reports, answers, research outputs, or documents ("which of these is better?", "evaluate these two responses", "A/B test these reports", "grade these two drafts"), even if they don't mention HTML, scoring, or the word "report".

Compress and optimize AI instructions to be clearer, more concise, and more efficient.

Wire an existing JSON Schema into VS Code, Cursor, and/or Zed so it provides validation and IntelliSense for matching JSON or YAML files.

Create a decision-complete implementation plan that can be executed mechanically without further design decisions.

Create or update an AI-optimized reference document for a tool, library, framework, or API via thorough research.

Create a PRD / product requirements document with app summary, design principles, and detailed requirements.

Create a self-contained prompt for ChatGPT Deep Research. Use when asked to prepare a research prompt, brief, or handoff from Codex to ChatGPT using the request and relevant local context. Does not conduct research or implement findings.

Create or iterate on a software design via a multi-agent workflow with parallel research and review phases. Stores artifacts in memory-bank/planning/{name}/.

Write a concise explainer / learning document / beginner guide that teaches a topic to someone unfamiliar with it.

Use when picking or vetting a keyboard shortcut on macOS. Triggers include "what hotkey should I use for X", "is this combo available", "does this shortcut conflict", "recommend a keybinding", "check this combo against my setup", or any mention of choosing, binding, or changing a shortcut in WezTerm, tmux, Zed, Chrome, Claude Code, or macOS. Determines whether a proposed combo collides with OS-reserved bindings, app defaults, or the user's customizations, and recommends ergonomic alternatives when needed.

Improve codebase architecture by investigating tight coupling, misplaced domain ownership, dependency cycles, or excessive compile/test impact. Use for module boundary audits, behavior-preserving structural refactors, and feedback-cost investigations driven by dependencies.

Initialize DESIGN-AGENTS.md config for the /design workflow, populated from project context.

Trace-based triage of web performance issues via chrome-devtools-mcp: LCP/INP regressions, slow page loads and interactions, ForcedReflow/layout thrashing, CLS, bundle bloat, memory leaks. Captures Chrome performance traces and analyzes them with performance insights. Requires setup-perf-stack. For installing react-scan, component-level React render attribution (unnecessary re-renders, unstable props), or automated render checks with Playwright, use react-scan instead.

Detect and remove dead code with knip. Use when the user asks to "run knip", "find unused files", "find unused exports", "find unused dependencies", "clean up dead code", "remove dead code", "set up knip", "configure knip", "knip.json", "knip false positive", "knip CI", or mentions a knip config, dependency bloat, bundle bloat from unused imports, or tree-shaking unused exports. Covers the configuration-first workflow, confidence-gated deletion, framework-specific gotchas (Next.js 15+, Tailwind, Storybook, Jest, Bun's test runner and bun build --compile), monorepos, CI integration, and performance tuning.

Generate LaTeX documents (or convert markdown to LaTeX) with professional typography. Optionally compile to PDF.

Compare AI model prices across Anthropic, OpenAI, and Google. Generates pages-per-dollar / cost-per-token visualizations.

Create a high-quality image generation prompt for Google's Nano Banana 2 (NB2) model on Gemini.

Optimize tsconfig.json for build speed, watch mode, and tsc memory. Use for slow tsc, too many file watchers, or TypeScript performance audits.

Comedy mode: talk in an over-the-top caricature of Claude Opus 5's much-mocked prose — load-bearing seams, genuinely, and this is where it gets interesting.

Install and use react-scan tooling for component-level React render attribution: "set up react-scan", "diagnose React re-renders", "find unnecessary renders", "find unstable props", "measure component renders programmatically", or automated render checks with Playwright (react-scan + playwright, lite headless API for CI). Covers install across Next.js/Vite/Remix/script-tag/browser-extension and the render-attribution → fix → validate loop. For trace-based triage of LCP/INP regressions, ForcedReflow, bundle bloat, or memory leaks, use investigate-web-perf instead.

Reflect on the conversation and propose improvements to durable AI-agent instruction files such as CLAUDE.md and AGENTS.md.

Render a change-review-guide as an interactive, self-contained HTML page (and optionally regenerated markdown) from a YAML content file, with the file index and stats pulled from git.

Save current session context to a handoff file so a future AI agent can continue with no prior context.

Bootstrap web performance debugging tooling: install chrome-devtools-mcp and related MCP servers/skills.

Add terms, context, or instructions to Shama voice-to-text files for clipboard mode (Ctrl+Opt+Space) or shell-command mode (Ctrl+Opt+S), at project or global level.

Initialize Shama voice-to-text config for the project — create voice.json, voice-vocabulary.md, voice-context.md, and register the project in Shama's registry.toml.

Manage software complexity through deep modules, information hiding, and strategic programming. Use when the user mentions "module design", "API too complex", "shallow class", "complexity budget", "strategic vs tactical", "deep module", "information leakage", "pass-through method", "this code is over-engineered", or "simplify this design". Also trigger when reviewing an interface for simplicity, evaluating whether an abstraction is pulling its weight, deciding whether a comment is worth writing, or choosing between general-purpose and special-purpose approaches. Covers deep vs shallow modules, red flags for complexity, and comments as design documentation. For code quality, see clean-code. For architecture boundaries, see clean-architecture.

Rewrite supplied text or the preceding assistant message in ASD-STE100 Simplified Technical English (STE).

Reflect on codebase navigation effectiveness at end of conversation. Surfaces dead ends, inefficiencies, missing context. Does not write files — pair with /kiro:steering-custom to persist.

Locate the on-disk Claude Code transcript file (.jsonl under ~/.claude/projects/) for the current or a specified conversation.

Use to thoroughly understand a software development objective before design & implementation: research, identify ambiguities, ask clarifying questions.

Stop. That last message did not land: re-pitch it.

Write automated tests, or decide whether code earns one. Covers mocking strategy (inject a test double before jest.mock), behavioral assertions, and code that resists testing.

Write or review agent instructions. Use when creating or editing skills, AGENTS.md, CLAUDE.md, system prompts, or reference documents agents consume.