Browse agent skills

Phase 1: The Refiner - Context Sanitization & Entropy Scoring

Motion films that never cut to the next slide: every beat grows out of the one before, one continuous camera, continuity measured by an oracle. A Claude Agent Skill for product launch films and feature demos.

Attach a structured decision log from an AI coding agent session to a pull request.

Terminal chat client for the AI assistant

On-device (and optionally cloud) ML inference for React Native: LLM, ASR, TTS, embeddings and vision, backed by the Xybrid Rust SDK

copycat - Claude Code skill: URL → pixel-perfect website clone. Playwright capture (3 viewports, folds, sections, hover), design-token extraction, pixel-diff verification loop. README EN/FR.

Search an artist, pick the tracks you want, and forge them into a normalised, segmented, manifest-backed audio dataset for training AI models.

Benchmark Apple Silicon memory bandwidth, disk speed, and MLX model inference performance to find the optimal LLM size

DSH Web GUI typography & readability tuner: wallpaper-aware contrast repair, edge recolor and hue-cycle effects, per-content rules (bold+underlined links, tinted headings) and a floating WYSIWYG preview window.

A lightweight and pluggable Telegram Bot API library for sending messages, managing chats, users, media, and bot interactions

The AI coding agent that actually does the work. Controls browsers, inspects your code, and writes tested software right inside your terminal.

Pi extension: the user types a secret into a masked dialog and applies it with a transparent ! command, so the agent never sees it

Diagnosis-first learning runtime for DeepSeek Harness — a live learning map, structured lessons, and persistent learner state, driven by the Universal Diagnostic Tutor skill.

Multi-turn autonomous agent trajectory evaluator measuring step redundancy and tool choice efficiency

Fault-tolerant tool execution circuit breaker detecting cascading upstream errors and triggering fallback routing

Long-horizon context window compression and semantic deduplicator eliminating redundant tool outputs

Local-first agentic tracing platform: inspect everything your agent did in a session.

M.E.K.S.U.S. CLI: see what your AI coding agents (Claude Code) and builds are doing, live, and get Discord alerts when they fail or need you. Status only: code, prompts and output never leave your machine.

Cognee persistent memory for the pi coding agent — native extension with auto-capture, auto-recall, and knowledge-graph tools. Fail-soft by design: pi keeps working when the cognee server is down.

Characters players can actually argue with: a persuasion mechanic for games where players type or speak to NPCs, judged by each character's own values. Powered by Jev.

中国建筑工程施工图图纸会审（自审）智能体技能：从一套施工图出发，系统找出各专业问题、跨专业矛盾与图纸集完整性缺陷，产出可上会签认的问题清单与会审记录表

Video-making skills for Claude Opus 5.5 in Claude Code — every frame and note generated in code, one skill per style: hand-painted animation, kinetic-typography reels | 用 Claude Opus 5.5 以代码制作视频的技能合集

OpenAI-compatible local router that benchmarks free coding models across providers and forwards requests to the best available model.

n8n nodes for UnuPlace: instant triggers and actions on leads, appointments, tasks, properties and orders.

n8n community node for Heymarket business texting

Claude paints with code: a Claude Code skill for procedural illustration in 59 styles, from ink wash (水墨), watercolour and ukiyo-e to stained glass, Roman mosaic, Art Deco, pop silkscreen, Delft tiles, thermal camera, X-ray, VHS tape, ballpoint doodles and a storybook animation. No image model: numpy + Pillow on a CPU, draw-on animation.

Connect your computer to ChatGPT, Claude and other MCP clients: npx remote-commander@latest remote

Verify that your automation actually did the work, inside your own process.

A resilient HTTP client with automatic queuing, retries, and bulkhead isolation

DSH bundle plugin with an isolated Mentor Agent tool, session-scoped history, and one global settings page.