Best Run AI Skills & MCP Servers

High-performance PDF processing MCP Server built with Rust

Skilder Runtime - NodeJS processes for edge execution connected to backend orchestrator via NATS

Walrus Memory MCP client — single-binary stdio MCP server that bridges Cursor / Claude Desktop / Antigravity / Claude Code to the Walrus Memory relayer. Handles browser-based wallet login on first run.

Local-first, LLM-agnostic memory layer for AI assistants. Run via npx @psraghuveer/memento.

TypeScript toolkit for Claude Code development — tool-integrated hook types with a runner plus a types generator for MCP tools and subagent state saving

A lightweight MCP (Model Context Protocol) client library for JavaScript and cross-runtime environments, supporting MCP Apps in host applications and multiple storage backends (Memory, File, Redis, Supabase, Neon).

A lightweight, zero-footprint TypeScript MCP server that lets AI assistants interact with the Godot 4.x game engine: not just editing files, but playing the game.

Read and publish to Substack via MCP — single Rust binary, no Node.js runtime.

Unofficial local-first Strava MCP server for AI training and performance agents.

Tap v2 plan format — public TypeScript types for the 11-op closed union, Plan discriminated union, and state-machine enums (Verdict, IntentState). Schema break vs v0.x; see ADR 2026-05-04-ecosystem-v2-launch.

FDKEY verification middleware for MCP servers — gate AI-agent access behind LLM-only puzzles. Runs on Node 18+, Cloudflare Workers, Bun, and Deno.

🐉 Wyrm — the memory layer that makes AI assistants remember, learn from failures, execute long-running goals, and federate across every session. MCP server with 126 tools across persistent memory, counter-pattern blocking, OODA agent loop, knowledge grap

MCP server for the Coalesce Transform API; run tools support Snowflake Key Pair and PAT auth

Core evaluation engine for MCPLab — agent adapters, MCP client, scenario runner, and result types

MCPLab - Test and evaluate MCP servers with LLMs — run evals, compare agents and launch the MCPLab web app

MCP server for Mockzilla (github.com/mockzilla/mockzilla-mcp). Lets Claude, Cursor, and Gemini CLI install the Mockzilla CLI, run local mock servers from OpenAPI specs, mock individual endpoints, and deploy hosted simulations on mockzilla.org.

IAPeer lifecycle primitive for ephemeral peers. MCP tool spawn(personality, runtime?, task, work_dir?) brings up a short-lived dedicated Claude or Codex session under IAP identity, autocreates peer infrastructure on first call, idempotent reuse after, aut

MCP server that exposes MCPLab evaluation tools — query runs, results, and traces via the Model Context Protocol

Universal MCP server for Hex Core — runs on Claude Code, Cursor, Continue, Gemini CLI, ChatGPT Desktop, and Zed. 13 tools over the component registry. Thin transport shell — pure-function builders live in @hex-core/payload.

Core library for parsing, loading, and managing .photon.ts files - runtime-agnostic foundation for building custom Photon runtimes

BotVault CLI — install the BotVault MCP server (Claude Code, OpenClaw, Cursor) so your AI agents can fetch credentials at runtime instead of you pasting them into chat.

Local and OpenAI embedding providers for documentation search — runs 100% offline by default

Runtime verification for AI coding agents. MCP server that lets Claude Code verify code changes work end-to-end via Playwright + OpenTelemetry + SigNoz.

MCP server for Afina anti-detect browser automation. Lets LLMs (Claude, GPT-4, Gemini, local) manage accounts, write/test/run RPA scripts & modules, schedule task groups, and inspect proxies/databases/variables in Afina via dialogue.