Browse agent skills

Open agent skills by Ian Balina (@diaryofamademan): playbooks AI agents can run end to end, starting with fundraising.

Evaluation suite for artkit kinds across models and schema profiles.

Vercel AI SDK 7 adapter for artkit: tools, approval, context, data parts and the useArtifacts hook.

React host for artkit artifacts: provider, view, editor, panel, card, approval, version diff, hooks and streaming preview.

🧠 Persistent Working Discipline, Shared Memory & Multi-Agent Handoff Skill for Claude Code, OpenAI Codex, Gemini CLI & Antigravity (Zero Dependencies, EN/VI)

Treat AI like a search engine: a Claude Code plugin that turns a prompt into a meta dump, an AI brain dump in one self-indexing JSON file of collected facts, keywords, anti-keywords, descriptions and sources that seeds any skill or subagent.

Universal engineering skill for AI coding agents. Routes each task to only the modules it needs (workflow, platform, stack, security) so agents plan first, verify what they claim, and spend fewer tokens on tool-call loops. Works with any Agent Skills-compatible client.

OSA Levenshtein + Meilisearch-style typo tolerance for FTS (ivf_levenshtein, transposition = 1), with an exact-recall IVF-Flat vector baseline

Claude skill: publication-quality ML / computer-vision architecture diagrams (3D CNN boxes, kernel cones, U-Net, FPN, ViT) from a JSON spec → SVG/PNG/PDF

An AI-powered advertising team for Indian small businesses. Research competitors, generate ad hooks, understand audiences and optimize Meta Ads on a small budget.

A Claude skill that reverse-engineers several apps, lets you pick the features you want, and builds one tested app from them. It checks compatibility, scans sources for security risks, and verifies test results against recorded runs. Standalone and MIT-licensed.

CAIT — a CAT workbench for DeepSeek Harness with translation memory, glossary, document tabs, MT routes, and Office/EPUB/subtitle/localization formats.

OpenAI instrumentation for Observe, emitting OpenTelemetry GenAI semantic conventions.

LangChain instrumentation for Observe, emitting OpenTelemetry GenAI semantic conventions.

passageway embedder for OpenAI-compatible embedding services: Ollama, LM Studio, vLLM, OpenAI, Voyage

TVSORG Chat SDK for React Native. Opens the TVSORG native chat on Android and iOS.

Claude Code skill for multi-agent runs: simple work goes to Sonnet or Haiku instead of Opus, to spend fewer tokens.

Claude Code skill that spends fewer tokens: the agent does the work and replies with one word.

Local music library for DeepSeek Harness: scan a folder, browse tracks with cover art and tags, and play through a sidebar-panel player with loop, shuffle and volume. · DSH 本地音乐插件：扫描音乐目录、读取封面与标签、侧栏入口 + 全屏播放器。

DocWorm SDK — document indexing, knowledge graphs, reasoning RAG, and chat

Ask Claude as usual, get an interactive page next to the chat: the answer up front, the options side by side, and a plan or a reply built from your picks. A skill for the Claude desktop app.

Analista de Instagram para o Claude: números reais via Composio MCP, por que seus posts bombaram, melhores horários, legendas, áudios em alta e concorrência. PT-BR.

Perplexity pplx-embed-v2-context contextual embeddings on Apple silicon with MLX

Contract testing framework for LLM prompts - PCSL (Prompt Contract Specification Language)

Understand the mechanism. Rebuild the behavior. Prove the result. Independent reverse-engineering MCP, skill and CLI for Claude Code and Codex: artifact analysis, experiment planning and behavioral verification.

remark plugin to support DenDen ruby annotations (`{漢字|かんじ}`)

Free, open-source alternative to Smootify and Storesynk (Shopyflow): Shopify products, cart and subscriptions in Webflow with HTML attributes. Optional Shopify to Webflow CMS sync.

Plugin-authoring skill for DeepSeek Harness: the distilled anatomy, client-UI recipe, slots, locale, theme tokens, install steps, gotchas, and validation checklist for building and publishing new DSH plugins.

🎧 DJ's Production RAG Pipeline - PDFs → Pinecone → LLM → RAGAS (Sub-10s E2E)

本地 Stable Diffusion 桥接插件：在「设置 → 本地绘图」里选一下 WebUI / Forge 目录就能用，注册 generate_image_local / generate_image_batch / sd_backend_status / sd_backend_control 四个工具，把本机 sdapi 后端直接接到模型手上。完全本地、零内容过滤、断网可用、后端按需启停。