Code indexing library for AI coding agents that parses source code with tree-sitter, chunks it into semantic units, and stores everything in a single SQLite file with vector and full-text search — no Docker, no server, no API keys required.
Cloudflare Workers-based MCP server that exposes Daiso product data through four tools: search_products (by query/category/maxPrice), find_stores (by lat/lng/radius), check_inventory (by store and product ID), and get_price_info. Serves over HTTP REST so it can be called from mobile apps.
Provides a secure, durable runtime for AI agent tasks using WebAssembly sandboxes with isolated execution, configurable resource limits, automatic retries, and lifecycle tracking for safe untrusted code execution.
A debuggable automation engine that traces every execution step and supports replay from any point, enabling browser automation, competitor intelligence gathering, and performance metrics extraction with full observability.
Universal domain availability engine written in Rust — CLI, library, and MCP server — with 1,200+ TLDs via IANA bootstrap, RDAP-first checks with WHOIS fallback, up to 100 concurrent lookups, pattern-based name generation, 11 curated presets (startup/tech/creative/finance/ecommerce), and JSON/CSV output with registrar and expiry info.
Local desktop sandbox for AI agents with full GUI support, providing MCP protocol-compatible cloud-style sandbox capabilities on your local machine with both shell and graphical interface.
SAGE (Sovereign Agent Governed Experience) — MCP/REST memory infrastructure giving AI agents BFT-consensus-validated institutional memory with confidence scores and natural decay; runs a real CometBFT node with 4 application validators (Sentinel, Dedup, Quality, Consistency) and an in-dashboard CEREBRUM neural-graph view, SQLite with optional AES-256-GCM, and multi-agent RBAC / federation.
Open-source desktop MCP proxy (Go + system tray UI for macOS Intel/Apple Silicon, Windows x64/ARM64, Linux x64/ARM64) that federates hundreds of MCP servers into a single `retrieve_tools` call — bypasses Cursor's 40-tool limit and OpenAI's 128-function cap, claims ~99% token reduction with 43% accuracy gain, and auto-quarantines new servers against Tool Poisoning Attacks until manual approval.
Bitget Agent Hub — a bundle of `bitget-mcp-server` (MCP Server for Claude Code/Cursor/Codex/any MCP-compatible AI), `bitget-client`/`bgc` CLI, and `bitget-skill` (for shell-based AI agents and OpenClaw) that lets AI assistants check prices, query balances, place/cancel orders, manage futures positions and leverage, and transfer funds on the Bitget exchange via natural language.
ClaudeR — an R package (MIT-licensed, PyPI `clauder-mcp`) that forges a direct MCP link between RStudio and Claude Desktop/Claude Code/Codex/Gemini CLI (or any MCP-compatible AI assistant including Cursor), letting the AI execute R code in the active RStudio environment and see results in real time for interactive coding and data analysis.
MCP server that exposes the OpenAI Codex CLI to Claude Code with session management and real-time progress — provides `exec`, `write`, `tdd`, and `review` tools supporting session resume.
MCP server that integrates Charles Proxy with MCP clients, enabling real-time packet capture reading, structured analysis, and low-token result returns.
Figma MCP server that streams live document data to AI tools without API rate limits, solving the free-tier limitation of other Figma MCP solutions.
Streamlines Telegram bot development by providing a robust, scalable Claude-powered messaging and interaction framework for Python projects.
**OpenClaw MCP Server** — a Model Context Protocol server that bridges AI clients (Claude.ai, Claude Desktop) with OpenClaw bots, enabling an AI-assistant-orchestrating-another-AI-assistant pattern where chat can delegate tasks to a Claw bot; deployable via Docker (GHCR) or `npx openclaw-mcp` with OAuth2 authentication support.
AI-driven semantic code search MCP tool powered by Windsurf's reverse-engineered SWE-grep protocol. Enables natural language queries to find relevant code files and line ranges across codebases, with bundled ripgrep and cross-platform support for macOS, Windows, and Linux.
**LinkedIn MCP Server** built with FastMCP + Patchright — search people/companies/jobs, scrape profiles with granular section selection (experience, education, contact info, posts, recommendations), and retrieve structured JSON from any MCP-compatible AI client; supports job search filters for date, type, experience level, and remote/hybrid/on-site.
Single-binary, zero-dependency MCP server for mocking seven protocols — HTTP, gRPC, GraphQL, WebSocket, MQTT, SSE, and SOAP — with import support from OpenAPI, Postman, WireMock, HAR, and cURL, plus built-in cloud tunnel sharing and chaos engineering.
An MCP server that enables AI assistants to perform packet analysis in Wireshark, supporting packet list summaries, frame details, TCP/HTTP stream following, and HTTP request and DNS query extraction.
MCP server for Xiaohongshu (Little Red Book) automation using Playwright. Supports auto-login, keyword search, note content retrieval, comment fetching, and intelligent comment publishing with multiple comment types including promotional, engagement, consulting, and professional modes.
Enables AI agents to programmatically search, analyze, extract insights, and interact with YouTube video content through intelligent API interactions.