Browse agent skills

TypeScript SDK for Redis Intent Router

A Claude Code skill for making game sprite frames with AI image tools: reference pictures and a design talk become a paste-ready prompts file, Codex or ChatGPT draws, and Python tools check, fit and review every frame until all are approved.

Workflow, house rules, references and mods so Claude Code stops shipping the AI-made look.

Skill per agenti AI (Claude, Codex, Antigravity) per la valutazione tecnica GDPR/ePrivacy sul trattamento reale: mappa dei trattamenti, matrice di controlli A–N, domande per DPO e legale. Non dichiara mai conformità. In italiano.

Skill per agenti AI (Claude, Codex, Antigravity) per assessment di sicurezza applicativa e cloud: regole di ingaggio, test dimostrabili, finding strutturati e copertura con denominatori reali. In italiano, per PA e piccole organizzazioni.

Official versioned Busymate AI SDK for the web: embed, end-user identity linking, the busymate-ai-v1 bridge and typed events

Универсальный AI-скилл для поиска работы: вакансии по городу, опыту и формату (офис/удалёнка/вахта). Работает в Claude, GPT, Codex, OpenClaw и т.д

OSSA toolbox: the deterministic half of a safety-first open-source contribution crew that runs on Claude Code.

A Claude skill that helps founders write a stronger YC application, check their founder video, and prep for the interview.

Make YouTube videos with Claude Code for $0: free guides + Claude Code skills (make-video, snipshorts, cartoonsmith)

A framework for decision-model harnesses: a decision model works through a task by answering the questions that code asks.

ZeroGPU provider for the AI SDK

Playwright logger for API and browser actions: colourful APIRequestContext logs with curl output, a per-test browser timeline, and secret redaction

Drop a file in a folder, get the work back with a signed receipt anyone can verify

AI skill that critiques website copy and page designs for conversion, backed by Nielsen Norman Group research and 80+ published A/B tests. Works with Claude, Codex, Cursor and more.

A typed HTTP client for zerolith over a neutral contract: generate the route table and types from an OpenAPI document, call operations with checked inputs, and read failures as RFC 9457 problems

澜姿：从姿势线稿和衣装集给社区鲸鱼娘大肥鱼出图，也可装进 DeepSeek Harness。Lanpose generates images of the community whale-girl Dafeiyu from pose sketches and costumes, and installs into DeepSeek Harness.

Gemini engine for the Form.io Universal Agent Gateway (UAG): Google's Gemini models, for structured extraction with research over read-only tools.

OpenAI engine for the Form.io Universal Agent Gateway (UAG): structured extraction, with research over read-only tools, on OpenAI or any server with its chat completions API (Ollama, vLLM, LM Studio, OpenRouter); and decisions with probabilities on OpenAI

Build engine packages for the Form.io Universal Agent Gateway (UAG): the engine contract, defineEngine, settings and instance resolution, the engine loader, and a conformance check.

AI agent skill that mentors hackathon teams from idea to submission. Works with Claude Code and OpenCode.

Keres SDK: browser automation plus runtime management client for Keres Browser. Connects to the local Keres Automation Gateway; Keres owns profiles, sessions, and the browser runtime.

Local, offline voice prompts for OpenCode, powered by Phonon-2.5

The agent loop that can't run away with your bill — budget guardrails, reliable structured output, typed tool calling, tenant-safe RAG. Zero runtime deps.

Connect Claude Code and Cursor to Plumb, and bring in your past chats.

LLM efficiency toolbelt: fit, tokenwaste, evalsanity, quantcheck, cheapest, modelcard — save money and catch silent bugs, on a laptop.

LangChain integration for llamay, a local LLM runtime

Samaya Finance Skills: finance research workflows for Claude, ChatGPT and Copilot, with every figure sourced through the Samaya MCP server

Open Claude skills for measuring and improving visibility in AI search, built on the AI Visibility Census method. CC BY 4.0.

Single-file Playwright test dashboard and HTML report: live progress, failures with screenshots, steps, console output and traces, re-run commands.