ai-skills

Create, verify, and improve AGENTS.md files. Minimal, focused, progressive disclosure. Fixes bloat, contradictions, stale info.

Audit and plan website optimisation for AI agents, AI search, LLM discoverability, llms.txt, structured data, and sitemaps.

Build distinctive, production-grade frontend interfaces (websites, components, dashboards, layouts) with polished UI design that avoids generic AI aesthetics.

Generate and validate Awesome list READMEs following sindresorhus/awesome standards.

Scaffold type-safe TypeScript projects with the Better-T-Stack CLI — new projects, features, or troubleshooting.

Build full-stack TypeScript apps with Convex — server functions, schema, auth, file storage, real-time, frontend integration, testing, deployment.

Exhaustive multi-agent code review with area-specific specialists, verified findings, and actionable remediation plans.

Create and customize GitHub profile READMEs with stats cards, animations, badges, and GitHub Actions automation.

Route design tasks to a competent model via opencode or claude CLI. Triggers on "design", "style", "UI", "CSS", "layout", "visual", "beautify", "redesign".

Create and edit content to sound authentically human, avoiding AI detection markers and generated writing patterns.

Guide for opencode CLI and web server — non-interactive mode, headless server, sessions, model config, and all CLI subcommands.

OpenRouter unified API for 400+ models — chat completions, streaming, tool calling, structured output, embeddings, multimodal.

OpenTUI project scaffolding, templates, examples, component library, and best practices.

OpenTUI with React — components, hooks (useKeyboard, useRenderer, useTimeline), JSX patterns, state, forms, testing.

OpenTUI core development with vanilla TypeScript imperative API — components, layout, input, styling, animations, debugging.

Multi-phase development with implementer/validator subagents, auto-retry, and automated build/test validation. For complex builds, not single-file tasks.

Collaborative multi-agent planning with specialist cross-review and iterative consensus-building through discussion rounds.

Deep multi-agent code review producing verified findings and fix plans. Triggers on "deep review", "security review", "code audit", "find problems".

Facilitate structured deliberation with multiple specialist subagents to reach consensus on complex decisions.

Generic tRPC implementation guide. Works with any framework (Next.js, Express, Fastify, Hono, Bun) and any package manager (pnpm, npm, yarn, bun).

Expert assistance for OpenTUI with SolidJS. Use for reactive components, signals, fine-grained reactivity, JSX patterns, and SolidJS-specific optimization.