MCP servers tagged llm
10K live MCP servers tagged "llm", ranked by trust score. Tags come from package metadata and repository topics, so this list covers servers across every category that work with llm.
Agenr V0
Local-first memory for AI agents. Extract, store, and recall knowledge from conversation transcripts.
Nex Code
The open-source agentic coding CLI. Focused on Ollama Cloud Models.
Evals
Evaluation framework for testing LLM agents' ability to use MCP tools
Backthread
Backthread keeps the thread on what your AI coding agent ships — it captures the why behind every change and turns it into a living 'How it works' view of your codebase you can ac…
Kravn
Self-hosted, open-source MCP gateway for enterprise AI governance and security.
Multi LLM Review
Parallel adversarial code review by multiple LLMs for Claude Code — Claude + Gemini by default, optional Codex/GPT for a 3-model panel.
TAGA 1
Hands-on reference implementation of an agentic AI stack that includes ReAct loop, MCP, hooks, streaming, token budgeting. A forkable learning baseline, not a production service.
Aftos Agent Stack
Self-hosted AI agent infrastructure: multiple autonomous agents running concurrently on local hardware, shipping production code with a 30-min/day human review loop.
Agent Manifest Specs
Open standard for declaring how AI agents can safely interact with web apps.
ZidoCode
Agent security & reliability tooling
AI Property Maintenance Advisor
Photograph an equipment data plate, get a cited service brief, or a straight answer that there is no reliable one. v2 is a LangGraph agent with memory, also served as an MCP serve…
LLM Agent Harness
TypeScript runtime turning local LLMs into controlled, observable agents (MCP, tools, traces) + document classification platform
LLM 101
从 0 开始，按照真实小白的认知路径，建立一张从 AI、LLM 到 RAG、Agent、MCP、Coding Agent，再到 AI 产业与硬件的完整知识地图。
Kollectiv
Plan, dispatch and review work across pooled AI workers, with free shared storage (Cloudflare R2/TeraBox), GitHub sync and connectors. Self-hosted, open source, no telemetry.
True Hands
An AI that really operates the computer — and can prove it. Built-in desktop driver, UI-only enforcement, project-scoped policy.
Treechats
Precise context control for AI. Treechats makes every model request inspectable, editable and versioned: see exactly what the model receives, change any part of it, and trace ever…
MNEMOS
Self-editing, bi-temporal memory runtime for AI agents — 6-signal hybrid retrieval, Neo4j knowledge graph, FSRS cognitive decay, sleep consolidation. Not another vector DB wrapper.
SequentialThinking
Multi-step sequential thinking MCP server powered by large language models — structured reasoning with self-MoA and iterative thinking modes
LibreLex IT
AI copilot for LibreOffice Writer: Italian legal drafting grounded on official sources. Citation verification, case-law research and template-guided drafting through mcp-legal-it.
Revoco
Undo for AI agent actions. Plans the rollback before the action runs, proves it still works, and rolls back a compromised grant's whole blast radius in one call.
Pekka Bot
Open-source AI workspace where bots do the work on their own cloud computers. A Chief of Staff bot delegates to specialist bots with memory, skills, schedules and plugins. Self-ho…
Youtube Digest Videos MCP
MCP server that turns YouTube videos into full text via Gemini (URL video understanding), with persistent SQLite cache and full-text search
Toolbox
Give your AI agent new tools without giving it your API keys. Sandboxed Lua commands, a key vault, and an MCP server built in.
Hookstack
Put agents on your production signals without handing them the keys. A pipe signs, routes and prices every signal; an investigator runs read-only and proves it at startup; what su…
Gaggle
Self-hosted, local-first AI agent workspace for macOS and Linux. Written in Rust with a single global SQLite database.
Marv
A terminal coding agent with a sandbox, subagents, MCP and any model (OpenRouter or Ollama). TypeScript on Bun.
Engram
Transparent memory sidecar for local and cloud LLMs. ACT-R activation scoring, provenance, ~90ms recall, one SQLite file, zero client changes.
AI Image Understanding Content Matching Engine
An end-to-end AI-powered relevance and visual content-matching engine with MCP and RAG capabilities, designed to evaluate, rank, and pair imagery with textual context using multim…
Portal
Manage 100+ providers, APIs, models, memory engines through one skill/mcp/API
Trinity
Trinity: A Triune Architecture for AGI Long-Term Memory
Halo Agent Showcase
HALO Agent OS: a hosted operating system for a one-person business, staffed by AI agents. Showcase; source is private.
Diagram First Agent
Agent 学习手册：一套以架构图为主线的中文 Agent 教程，它先给出 Agent 平台的分层总览，再沿八级演进逐层展开：ReAct → 工具调用 → 模型网关 → 工具网关 → RAG → 意图识别 → 多 Agent → 平台化。每一章以架构图开篇，配读图步骤、请求时序、故障模式与优化矩阵，并补充内容安全、隐私、合规、测试、发布、长任务可靠性等生产…
Tbelt Code Public
Desktop coding agent that works with any model provider and runs commands in your operating system's sandbox.
0x Sneaky Alpha
0x Sneaky Alpha: bring-your-own-key AI chat for iPhone (PWA). GPL-3.0.
Neosian
The state layer for LLM agents: durable conversations, agent-curated memory, skills, a shared board and a ledger of who did what — on storage you own.
Willdeep CLI
自托管 AI Coding Agent（Rust）：常驻 Runtime 撑起终端 / 浏览器 / 手机三端；每行改动可追溯到具体 Agent 与工具调用，子 Agent 的活由退出码裁决，检查点回退、目标 token 预算与验收清单跨轮次跨重启累计，用量与反馈账本留在本地。接任意 OpenAI 兼容或 Anthropic 端点。Self-hosted a…
Checkride
Static scanner that finds the dangerous things an AI agent can do without asking a human first, for MCP servers and agent tool code.
Comodor
The Agent. Many Algorithms. Your Way.
Curatex MCP
Agentic MCP content pipeline: Reddit/YouTube/News → script, voice, and video
Ranite
General-purpose AI agent for the terminal: local tools, MCP servers, Agent Skills, and durable memory, with workspace-scoped permissions and one long-running custodian session.
Control Tower
Your own AI team, on your own server. Self-hosted AI agents that watch your product, report what matters and suggest what to do — nothing happens until you say yes.
Oss
HUMMBL open-source monorepo — governance primitives for AI agent fleets: kill switches, circuit breakers, cost governors, coordination bus, 120 mental models (Base120). 46 Python…
Toolshape
Convert LLM tool definitions between OpenAI, Anthropic, Gemini and MCP formats, with the schema fixes each one needs.
Grokbot
LocalCrew: an independent, self-hosted AI teammate platform (local Ollama or any OpenAI-compatible API). Not affiliated with Grok or xAI.
ApiWall
Every public API on one zoomable wall. Explore 2,150+ APIs, 20 MCP servers and a curated AI Models list. Search, filter by category or auth, get starter code, and open the docs in…
Claude API Lab
Interactive desktop workbench & wire inspector for the Anthropic Claude API (Messages, SSE Streaming, Tool Use, Adaptive Thinking, Prompt Caching, and Custom MCP Servers)
Mcpxx
Header-only C++20 implementation of the Model Context Protocol (MCP 2026-07-28) – dual transport (stdio + Streamable HTTP), easy tool servers, ready for llama.cpp / llama-ui
Berghain MCP
BERGHAIN MCP server — sovereign AI stress-testing
Google Workspace MCP
Mono-tool MCP server for Google Calendar, Gmail & Drive — optimized for LLM context efficiency
Two Layer Agent Contract
The Two-Layer Agent Contract: guidance influences candidate behaviour, runtime policy enforces executable authority. Reference implementation for 'Intent Is Not Policy' talk.