Live file-activity graph for opencode sidebar (which files the agent is reading / modifying)
Directory
Browse agent skills
Search by intent: design systems, testing, deployment, security, documentation, Azure, Supabase, React, or any workflow you want an agent to perform better.
Showing 6,541–6,570 of 125,958 skills
Page 219 of 4,199 · loading 30 at a time
llama-index protocols AG-UI integration
n8n node: redact European personal data (names, EU national IDs, IBANs, emails, phones) before it reaches an LLM, and restore it afterwards.
Mobilint NPU integrations for Hugging Face Transformers: LLMs, VLMs, ASR, and speculative decoding
Claude Code-style /context for pi: see what fills your context window, by category.
LangChain integration for the Decodo web scraping API
Store-compliant, call-style reminders for React Native: a full-screen incoming-call screen on Android, Time Sensitive notifications on iOS, and spoken reminders via the device text-to-speech engine.
DeepSeek Harness (dsh) plugin: Decodo Web Scraping API as the provider behind the built-in web_fetch tool, via @decodo/sdk-ts.
ElJay AI — a free, local, terminal-based universal AI assistant with Ollama LLM, web research, image & video generation, skills, knowledge base, and project-local memory. Standard library Python only.
Coding agents fill the blanks in a prompt and ship the guess. For Codex, Claude, Cursor, Antigravity, Gemini, and Grok: Karpathy's coding rules, plus a short gap check. Ask only when a missing fact would change the result and you cannot look it up.
Native body tracking, pose rules, exercise counting and body interaction for React Native and Expo. Camera, inference, tracking and rendering run natively.
<p align="center"><img src="https://raw.githubusercontent.com/krishnapollu/playwright-scout/main/docs/img/logo.svg" alt="Scout logo" width="128"></p>
Static indexer and query library for Playwright test projects. It parses source without running tests or evaluating project code.
AI Coding Token Economizer & Model Fatigue Shield for Cursor, Claude Code, Windsurf & VS Code
Paidwen SDK: the agent asks Paidwen before each action and applies the answer.
Safe, idempotent webhook handling for Node.js and TypeScript
Cowitness: share a moment of your day and the people you share with witness it, one at a time, and can answer typed or spoken. Rules, a Firestore store, Next.js route handlers and React screens, for any app that says who the people are and where things ar
Crash reporting (Sentry), session replay (LogRocket), one captureError/captureWarning seam and a root ErrorBoundary for TilePacket apps. Real on iOS/Android release builds, console-only on web and in development.
Research assistant with diagnostics, recording and declarative browser assertions
Per-channel Meilisearch indexes, suggest popup, and optional LLM-augmented search.
A local, self-contained Playwright report with run history and flaky-test tracking
Attention Span output styles (Attention-kind, Spartan, Rundown) plus /tldr, packaged for the Pi coding agent.
DSH小助手:原生 Swift 悬浮窗负责展示与交互,宿主插件负责把它接到 DSH 会话状态(含今日工作总结)
RAWAN: Autonomous Multi-Agent Swarm Intelligence & Self-Evolving Codebase Fortification Engine
A persistent, Letta-backed Vercel AI SDK Agent with memory, dreaming, application-owned tools and human-in-the-loop.
dsh 工作区代码审查插件:把 open-code-review(ocr CLI)封装为 5 个模型工具 + 设置卡片(从 settings.yaml 选 provider/model 应用到 OCR 配置)。
dsh 工作区语义检索插件(zg):zg_search / zg_index / zg_status + 本地 embedding 设置卡片 + systemPrompt 路由规则 + guard 安全兜底。
Live tokens-per-second (TPS) + time-to-first-token (TTFT) meter for pi, tuned for local models (footer status line + /tps command).
Bot and abuse protection for AI-powered applications. Server-side request admission for Node.js.