Browse agent skills

Cursor-aware reconnecting chat transport for Vercel AI SDK

The Wilma assistant as a themed React widget on @wildix/uikit: compound components, default leaves and the DOM platform.

Project Driver for cross-Agent project continuity and progression; execution, evidence, verification, and recovery are the underlying trust substrate.

Agent-first unofficial Gemini Notebook client for Node.js — JSON CLI, citations, multi-turn chat, all studio artifacts, research & notes

TypeScript reimplementation of the Amiga narrator.device and translator.library

Capable is a TypeScript framework for defining a capability once and exposing it through the surfaces your application needs: typed calls, HTTP, CLI, MCP, AI tools, and more.

Streaming Gemini provider for pi, built on the agy CLI. Registers antigravity/* models in pi's /model picker via SQLite polling + protobuf decode of agy's conversation DBs.

Embedded HelixDB runtime and native graph algorithms for the HelixDB JavaScript SDK

Material black theme for the pi coding agent - near-black UI, green user messages, and a highlighted final answer so you can see where a run ended at a glance.

Gera documentos de ctx-builder para IA (ctx-builder do Marvin e Spec-Driven Development) dentro de um projeto.

Translate an Amazon Bedrock Agent InvokeAgent response stream into an AI SDK UI message stream, for use with useChat.

A minimal, self hosted OpenAI compatible adapter for one shot completions through locally authenticated CLI model tools. A development and testing bridge, not a hosted model service.

Framework-aware automation CLI: scaffolds and runs clean/build/dev scripts per project type, an environment doctor, a SwiftPM-aware iOS build engine, and a git workflow helper. Runs natively on macOS, Linux, and Windows.

Knowledge graph engine for understanding TypeScript/JavaScript codebases.

CLI tool for deploying static sites to EdgeOne with automatic domain, CNAME, and SSL configuration

Fast, provider-agnostic AI coding assistant — BYOK, any provider, all from your terminal.

LLM-powered judge for opencode - blocks file modifications without explicit user permission

Guided assistance for web journeys. An assistant sees the structure of the page, never its contents, and highlights the next step for the customer.

A draggable desktop pet that reacts to coding agent task states.

Continuous verification of agent skills: run a skill's eval suite with and without the skill across model versions, emit signed dated receipts, and diff receipts into drift reports.

Browser-based OCR SDK powered by PaddleOCR, ONNX Runtime Web and OpenCV.js

Accessible, headless, Native CSS-first UI primitives

Official TypeScript SDK for the Slide API

Embedded graph-first multi-model database — Node.js & React Native binding

Production-ready Android Engineering Rules for Cursor and other AI coding assistants. Build consistent, scalable, and maintainable Android apps with Kotlin, MVI, Clean Architecture, XML, and Koin.

A linter for your LLM token spend. Paste your agent's trace, see exactly where you're burning tokens — and how to stop. Runs 100% in your browser, nothing uploaded.

Management & supervised self-improvement client for Ratel Cloud — catalog writes, intent analysis, and skill-suggestion review over the cloud API, plus the OpenTelemetry destination on /otel. Dependency-free at the root; pure HTTP, no native addon.

Literature-fetching & citation-auditing skill for LLM agents (Claude/Codex) — 讓 AI 引用經得起追問:找文獻、查核引用真偽與支持度、文獻支撐寫作,防幻覺引用。

Use Codex, TraeX, and compatible coding agents from Lark/Feishu.

Vehicle's runtime-neutral wire contract: operation descriptors, schema codecs, failure shapes. Zero runtime dependencies, zero Bun-specific code -- the one thing every Vehicle client and server package depends on.