Browse agent skills

Shared ML inference framework for camera.ui detection plugins (onnx, openvino, coreml, ncnn)

Local-first shared memory layer for AI coding agents

Your personal AI CLI — free, fast, and smart

WebSocket HTTP/WS reverse tunnel client — embeddable library (createTunnel) + CLI + a typed relay SDK (VolterClient).

VibeDeploy by Denkprozesse — deploy your vibe-coded apps to EU/GDPR-safe hosting straight from your AI tool (Claude, Cursor, Windsurf …) via MCP.

Detect whether a JavaScript CLI is running inside a coding agent.

Minimalist AI CLI assistant — one shell tool, filesystem memory, auto context migration

No description available

Scaffold de docs para projetos TypeScript com desenvolvimento assistido por AI: gera CLAUDE.md + docs/ a partir de templates vivos (common + perfil cli/ssr).

HuggingFace Inference API provider adapter for Anchor Execute

The official Node.js REST API wrapper for Rymi.

PREVIEW (not part of the supported v1 content seam) — build-time embedding generation for semantic search. A fresh build environment downloads the all-MiniLM-L6-v2 model (~25MB via @huggingface/transformers) on first run; cache the model dir in CI. Experi

Declarative speech-recognition component for the AmiVoice cloud API. Server-side APPKEY, browser mic capture, two-channel relay, via wc-bindable-protocol.

libSQL / Turso source adapter for @mirk/store's async ports. KV + collection store with atomic writes and transactions, and a native-vector similarity store, over one libSQL connection (remote, file:, or :memory:).

Zero-dependency OpenMetrics/Prometheus exporter for Bun — Counter, Gauge, Histogram, /metrics HTTP endpoint, process-level collectors

Shared OpenF1 schemas, query builder, and OpenAPI generation

Claude skill for Aprimo references (<ref:…> DSL): docs + a bundled compiler to execute, lint, and debug them.

System-wide AI overlay. Hidden from screen sharing.

Elite Claude Code skill for full system architecture & design

Declare a UI component once — its locator, typed elements, and actions — then `mount()` it into a fully-typed Playwright test handle.

Scaffold a new OpenCanva project — `npm create opencanva@latest`.

Project Bootstrapper (#5) engine — turn a spec (template + feature toggles) into a complete, production-ready, AI-agent-ready project (code + CLAUDE.md/AGENTS.md/mcp.json + CI + Docker) as a file map. Node native TypeScript; zero-network.

System Design Assistant (#6) engine — generate architecture diagrams (Mermaid), ADRs, weighted tradeoff analysis, persisted design memory, and interview-prep grading. Node native TypeScript; zero-network.

Agent Monitoring (#4) engine — 'Datadog for agents'. Record agent runs (reason→llm→tool trajectories) via a framework-agnostic SDK, then get cost-per-task, latency, error rate, model mix, failure taxonomy, and alerts. Node native TypeScript; zero-network.

Codebase Intelligence (#2) engine — RAG over code. Ingest -> chunk -> embed -> hybrid retrieve (vector + lexical) -> grounded Q&A with citations. Runs on Node native TypeScript; zero-network by default (local embeddings), real embedding/LLM APIs swappable

Openbuff CLI - AI Coding Assistant with NVIDIA NIM

The workflow framework for AI-assisted development. Four operating modes, 25 golden rules, and anti-hallucination guardrails for 9 frameworks.

Open-source smart LLM router for the Omnicall gateway. `npx omnicall-router` works FREE out of the box — point any OpenAI client at it, send model "auto", and it routes to the cheapest of 248 capable models. Pass a funded gck_ key to unlock all 248 pay-pe

AXI-compliant task/backlog CLI — token-efficient TOON output, pluggable backends, byte-exact markdown round-trip, idempotent mutations

Contract-first Copilot scaffolding and workspace bootstrap scripts for Edenred repositories.