Packages

MCP server that enables Claude Desktop to perform real-time web research via Google search integration and webpage content extraction using Playwright. Provides search, page visiting with content extraction, screenshot capture, and research session tracking.

US federal government REST API (CASE Search API) providing programmatic access to federal district, bankruptcy, and appellate court dockets and documents via the Public Access to Court Electronic Records system.

Open-source data catalog that helps teams discover and manage data assets across an organization. Ships as a single binary and catalogs databases, APIs, message queues, and data pipelines with full-text search, interactive lineage graphs, and metadata management.

Asynchronous enumeration and vulnerability scanner that uses Celery task queues to orchestrate network scanning tools (Nmap, Nikto, Gobuster, sqlmap, etc.). Supports service-aware scanning, job control (pause/resume/cancel), subdomain enumeration, and HTML report generation.

Curated directory of Web3-focused MCP servers organized into six categories: Chain-RPC (blockchain interaction), Trading (DEX/CEX), DeFi (liquidity/yields), Market Data (pricing/analytics), Tool (NFTs/smart contracts), and Social (Discord/Telegram). Also promotes DeMCP, a decentralized MCP network with TEE and blockchain-based security.

Open-source vision-language models (GLM-4.6V, GLM-4.5V, GLM-4.1V-Thinking) for multimodal reasoning including image/video/document understanding, visual function calling, interleaved image-text generation, and GUI agent operations.

Standardized format and CLI for packaging and distributing local MCP servers as .mcpb bundles (zip archives with manifest). Enables one-click installation of MCP servers in desktop apps like Claude, similar to Chrome/VS Code extensions.

Ultra-lightweight personal AI assistant distributed as a single static Go binary with built-in semantic memory (SQLite + sqlite-vec), multi-provider LLM support, chat app integrations (Telegram, Discord, Slack, WhatsApp), scheduled tasks, and multimodal input handling.

An Emacs-native AI assistant for agentic development that uses LLM reasoning to select and execute tools (read/write files, run commands, list files) within the Emacs workflow.

Memory operating system for LLMs and AI agents that unifies store, retrieve, and manage operations for long-term memory, enabling context-aware and personalized interactions with multi-modal support.

Open-source low-code platform for building internal tools, admin panels, and dashboards by connecting to databases and APIs through a visual GUI builder with JavaScript widget customization. Provides a management REST API for programmatic control of the Appsmith instance itself.

The official Chrome Status Dashboard that tracks feature launches in the Blink browser engine. Guides feature owners through the Chrome launch process and serves as a primary source for developer information about web platform features.

Open-source enterprise AI agent platform that combines RAG (retrieval-augmented generation) with agentic workflows and MCP tool-use support. Enables document upload, web crawling, automatic text segmentation and vectorization, and zero-code integration into existing business systems for intelligent Q&A. Supports multiple LLM backends including DeepSeek, Llama, Qwen, OpenAI, Claude, and Gemini.

Agent framework designed for shared environments with explicit boundaries and visible execution evidence. Uses a hook-first architecture built on pluggy, with skill discovery, session event logging, and comma-command mode for internal operations.

Curated catalog and registry of open-source MCP servers maintained by Metorial. Lists official vendor and community-built MCP servers across categories including web search, cloud platforms, databases, APIs, and dev tools. Metorial also offers a serverless MCP hosting platform for single-line integration of listed servers.

A local-first personal AI assistant built on elizaOS with desktop and CLI interfaces, supporting multiple LLM providers (Claude, GPT, Ollama) and platform integrations (Telegram, Discord).

Solution template combining LangGraph with MCP to create a Universal Assistant that routes user messages to appropriate agents and MCP servers using vector-database-backed intelligent routing.

Research implementation (NeurIPS 2022) for improving tabular deep learning by transforming scalar numerical features into vector embeddings before feeding them into neural network backbones like MLPs and Transformers.

A lightweight Kotlin/JVM library for calculating Islamic prayer times and triggering the Azan (call to prayer), supporting 10 different calculation methods from organizations like Umm al-Qura, Muslim World League, and ISNA.

Lightweight LLM inference engine in pure C11 that runs a 1-billion parameter model on minimal hardware ($10 board, 256MB RAM). Zero external dependencies, ~80KB binary, ~45MB runtime RAM. Supports GGUF models with quantization, SIMD optimization (ARM NEON, x86 SSE2), grammar-constrained JSON output, and KV cache persistence.