Browse agent skills

Official JavaScript/TypeScript SDK for Wan 3.0 multimodal AI video generator, text-to-video, and document-to-video studio.

Typed decisions in the browser: run the open-jev model (a Jev-shaped System One model) with Transformers.js.

Session-tree token usage, live performance, context window and estimated cost for OpenCode 2

Demo npm distribution for the cann-log-analysis Skill group.

Trellis workflow patch installer for Pi, Cursor, Claude Code, OpenCode and Codex.

nx-rp（npx-repo）—— 外部信息以链接方式管理：link / doc / workflow 三个域，CLI + Web 面板同源，按 cwd 自动隔离 scope

Dual-mode CLI for coding agents: skill management + headless task execution + the workflow engine (dual-executor: skills in-session, CLI headless — ADR-395).

Convert timestamped text files to MP3 using Mistral Voxtral TTS CLI

Official JavaScript/TypeScript SDK for MiniMax H3 multimodal AI video generator with native stereo audio and 2K cinematic output.

Provider-agnostic LLM observability & reliability SDK — semantic caching, retries, rate limiting, cost tracking, and a local dashboard via a single @track() decorator.

Official JavaScript/TypeScript SDK for Gemini Omni multimodal AI video generator and conversational video editor.

Javascript library to display knowhy chatbot on your website

Official JavaScript/TypeScript SDK for Seedance 2.5 native 30s AI video generator with multimodal references and synced audio.

Personal paste inbox for DeepSeek Harness (dsh): capture, classify, browse and retrieve.

Skill2CN · 技能汉化 — translate DSH skill descriptions to Simplified Chinese

Pi extension that reads completed assistant responses aloud with OpenAI text-to-speech and mpv

Content-aware shell command timeouts for Pi: caps every bash/powershell call so a hung command cannot stall the agent, while long-running builds keep the room they need.

SealedRun Record verification: canonicalization, hash chain, hybrid signatures

Browser agent that acts in ~0.5s per step via Jev. Headless-ready, works on your OpenRouter key. Python env bootstrapped on first run.

Ask an AI assistant from a MeshCore: /ask <question>, answered by an OpenAI-compatible model.

给同事的运维请求通道：同事在飞书私聊机器人提一件事，你在审批卡上点「同意」后，由你指定的那条 DSH 会话带着你配好的工具与人设去办，结果以卡片回到提交人的飞书（Markdown 正常渲染），不需要公网入口。四个特点：配对码把发送者绑成管理员，机器人只服务指定账号；非管理员请求走审批卡片，同意/拒绝都校验点击人身份；请求串行排队，前一条整轮 turn 跑完才注入下一条；目标会话启动时主动 resume，不用先有人在界面上点开。

What your language costs you in tokens. Measures how many more tokens the same sentence takes in Turkish and other languages than in English, across the tokenizers real models use.

Text-in, theme-crossed lexical field analysis out.

n8n community node for Cactus Needle 3 — on-device tool-calling via your cactus-docker playground server (POST /complete, /reset, /model, /load-model, /finetune).

A modern, offline-first content management platform built with Nuxt 4, featuring page and email builder, file and data management, and AI-powered document processing

Local browser OCR with verified PP-OCRv6 models and a SIMD WASM worker

Batteries-included local multilingual MoE Text-to-Speech with acoustic classifier, audible calibration testing, router-in-the-weights, and calibrated LoRA personas for Pi

Alibaba Cloud Elasticsearch Enterprise Knowledge Engine integration for OpenClaw, Hermes, Codex and Claude Code

OpenCode V2 plugin: xAI SuperGrok web search, X search, Grok Imagine, speech-to-text, and text-to-speech.

Classify Hugging Face datasets across typed semantic dimensions with TypeSafe Jev System One