MCP servers that ship as code — a repository or a package you install and run locally (usually over stdio). They have no hosted endpoint, so we cannot probe or grade them: instead each one shows its repository, install command, license, stars and when it was last updated.
Collected from GitHub repositories and npm, PyPI, Cargo, Go, Maven, NuGet and RubyGems packages tagged as MCP servers, the official MCP Registry, Docker's MCP catalog, ModelScope, Baidu MCP World, Cline and other public directories; one entry per repository. Each candidate is checked by a language model from its name, description and topics, and only actual MCP servers are listed (not clients, SDKs, lists or skills). Package descriptions, licences and links are read from each package's own registry (npm, PyPI, crates.io, NuGet, RubyGems; Go and Maven packages are only checked to exist), and packages the registry no longer has are removed. A server that also offers a hosted endpoint is listed under Remote instead. These entries are not counted in our reachability, verification or readiness statistics.
-
Experimental always-on runtime for AI coding agents: 5-tier memory, pattern learning, MCP server, Telegram + webchat.
MIT -
squish-memorygithub.com/michielhdoteth/squishupdated this year
Local-first memory runtime for Claude Code, Codex, ChatGPT, and AI agents. Searchable session history, durable memory, and cross-tool sync. No external LLM required.
npm MIT -
SEAM: local-first memory runtime for AI agents, with retrieval and glassbox provenance over MCP.
pypi -
Runtime for a persistent autonomous research agent on Claude: scheduler, MCP tool server, memory index, safety hooks, change control
-
bedrock-agentcore-strands-mcp-runtimegithub.com/Kjeff24/bedrock-agentcore-strands-mcp-runtimeupdated this year
Production-ready AWS Bedrock AgentCore runtime using Strands agents with Atlassian MCP tool integration, Cognito-authenticated streaming chat frontend, and Terraform infrastructure for ECR, Gateway, Memory, and Cognito.
-
Codememorygithub.com/byte271/Codememoryupdated this year
Runtime behavior memory for AI-generated code — captures intent, execution traces, and failures via MCP, then returns repair briefs so agents stop guessing.
MIT ★ 2 -
Squish gives AI agents calibrated recall confidence (ECE 0.055) and honest abstention — the only memory runtime that tells your agent how much to trust each recollection, and says I don't know when nothing clears the bar. Local-first, MIT, npm install in 30s.
MIT ★ 12 -
Metadata-only PyPI and MCP Registry distribution for the Qdrant Memory Runtime
npm MIT ★ 1 -
pargithub.com/latentfidelity/parupdated this year
Persistent Agent Runtime — self-hosted cognitive runtime for AI coding agents: semantic memory, knowledge graph, agent coordination, event-driven workflows, and autonomous self-maintenance via MCP
MIT ★ 9 -
Official Lucy MCP runtime for governed agent memory. Three tools only: recall, propose, proposal status.
npm Apache-2.0 -
statepilotgithub.com/newsbruno/statepilotupdated this year
Predictable browser runtime for AI agents with memory, validation, recovery, and MCP integration.
MIT ★ 1 -
Persistent, cross-runtime, typed memory for AI agents — MCP stdio server over embedded SQLite. Local, offline, no cloud.
MIT -
LifeOS — AI-native knowledge OS runtime, CLI scaffold, and MCP memory server
npm MIT -
memdgithub.com/robll-v1/memdupdated this year
Local-first memory runtime for AI agents with SQLite, REST, and MCP.
Apache-2.0 ★ 2 -
personal-modelgithub.com/intuition-lab/personal-modelupdated this year
Local-first macOS Runtime for screen-context memory, personal modeling, and MCP.
pypi Apache-2.0 -
TinyBuddy — a full-stack, mobile-first AI teammate platform: agent runtime, memory, skills, knowledge base, MCP (client + server), cron scheduling and RBAC.
MIT -
cclggithub.com/schift-io/cclgupdated this year
Local-first memory, hook, MCP, and code graph runtime for coding agents
pypi MIT -
Local-first, self-hosted AI agent runtime and MCP bridge with sandboxed sessions, memory, credentials, audit/replay, and a local Console.
Apache-2.0 ★ 643 -
Source-grounded MCP memory runtime and Obsidian-compatible vault template for LLM agents.
MIT ★ 76 -
A growable, time-aware memory graph for AI agents. SQLite by default (zero runtime dependencies), Postgres optionally, configurable embeddings. JS API, CLI, MCP server, HTTP server.
npm MIT -
Cursor integration for IronMem: local stdio MCP, project recall, and explicit durable memory. Runtime installed separately.
Apache-2.0 -
brain-mcpgithub.com/ugur-murat-alt/brain-mcpupdated this year
Zero-trust, local-first MCP server and explainable AI memory runtime for Markdown knowledge bases
Apache-2.0 -
Kaleidoscope: a local memory runtime for coding agents.
npm SEE LICENSE IN LICENSE.txt -
Long-term memory for coding agents that you own: Markdown on disk, a disposable SQLite index, zero runtime dependencies.
MIT -
MCP server for AnvaiOps — give any AI agent operational memory: provenance-rich incident search and attributed memory writes. Zero runtime deps.
Apache-2.0 -
beinggithub.com/wnbhr/beingactive
Personality Runtime — persistent memory, identity, and relationships for AI agents that work across multiple LLM platforms via MCP
★ 1 -
@memoryostm/mcpupdated this year
MemoryOS MCP adapter for universal agent memory across MCP-compatible runtimes.
npm MIT -
AXLR: Underpass's agentic execution runtime. A console, one-command runs and an mTLS HTTP/gRPC/MCP service, with a policy, budget and correlation ids per execution; KMP memory and MADE procedures over MCP.
Apache-2.0 -
causalos-mcpupdated this year
Dual Memory Agent Runtime: Causal Graph Engine + General Memory + Forward Simulation + Backward Inference for AI Agents
npm ISC -
A persistent, stateful runtime for Claude Code: memory, watchers, voice, and an optional robot body.
MIT ★ 4 -
kugithub.com/zhangshaoling/kuupdated this year
Ku/Dao executable-memory language runtime with C VM, self-hosting frontend, and MCP gateway
MIT ★ 1 -
One owner's memory MCP on Cloudflare Workers + D1, with a stateless ledger of who connected and which method they called from where. OAuth for claude.ai, static bearer for Claude Code. Zero runtime dependencies.
MIT -
Local-first memory, agent dispatch, and workflow runtime for AI coding agents, with shared context through MCP.
MIT ★ 1 -
File-first memory runtime for AI agents — versioned, portable, and always there when the next session starts.
MIT ★ 6 -
Memory patrol + lexical recall + conversation gates for long-lived agents. Auditable SQLite kernels, CLI + MCP, zero runtime deps. 记忆巡检、召回打分与群聊接话决策内核;可审计、零依赖。
MIT ★ 1 -
Local-first memory + output verification for AI agents. Zero runtime dependencies, works offline.
pypi AGPL-3.0 -
Local-first AI memory runtime with hybrid retrieval, temporal memory, privacy controls, MCP support, and explainable context generation.
-
membraingithub.com/quangdang46/membrainupdated this year
Brain-inspired, local-first memory runtime for AI agents with explainable retrieval, provenance-first storage, and policy-aware memory lifecycles
★ 2 -
Private local memory for any AI agent: Markdown database, SQLite search, MCP, zero runtime dependencies.
MIT ★ 9 -
QwenPaw plugin wiring agents to OpenViking long-term semantic memory: upstream's Node.js MCP proxy ported to native Python — 7 tools, 5 skills, no Node runtime, no third-party deps.
AGPL-3.0 -
Shared local memory runtime for AI clients and agents. Exposes CLI, HTTP API, and MCP above pluggable providers like mem0.
MIT -
Self-hosted shared memory, encrypted credentials, and scoped connectors for AI agents. MCP and HTTP for Claude Code, Codex, Cursor, and custom runtimes.
MIT -
Memory substrate and agent runtime for AI systems that have to be trusted — grounded facts and simulated hypotheses never mix, and every retrieval call reports index drift instead of hiding it.
Apache-2.0 ★ 3 -
An open protocol and Rust runtime for persistent, user-owned context and memory across AI applications. | 面向 AI 应用的持久上下文与记忆协议及 Rust 运行时,数据由用户持有。
MIT ★ 1 -
Local-first memory for AI agents. Structured synthetic memory cards (decision / preference / risk) with verifiable state history via ERC-8350. Offline, no account. MCP + 7 agent runtimes.
MIT ★ 12 -
codememgithub.com/cogniplex/codememupdated this year
A standalone Rust memory engine for AI coding assistants. Single binary, zero runtime deps. Graph-vector hybrid with MCP tools.
Apache-2.0 ★ 18 -
MCP server that gives AI agents evidence from a running Windows process: memory, runtime disassembly, IAT resolution, call-chain tracing, and a hypothesis ledger.
MIT -
exoprocactive
Unified CLI and Model Context Protocol (MCP) server for native Windows x64 reverse engineering, thread hijacking, function redirection, memory inspection, and runtime instrumentation.
npm MIT -
mcp-system-infogithub.com/thomaswillix/mcp-system-infoupdated this year
MCP server providing system information tools (memory, CPU, disk, Java runtime)
-
irontology-mcpgithub.com/PromptExecution/irontology-mcpupdated this year
irontology - a memory safe rust Inference Runtime Ontology
-
Restaurant booking chatbot on Zalo — custom MCP server as a separate AgentBase runtime, Private MCP Gateway, returning-guest memory, real Zalo Bot Platform webhook and web simulator.
MIT -
nexo-braingithub.com/wazionapps/nexoupdated this year
NEXO runtime core for NEXO Desktop. Provides local memory, Deep Sleep, Evolution support-ticket mode, skills, watchdog, and MCP tooling for the Desktop product.
npm AGPL-3.0 -
Persistent memory for Claude Code. Survives compaction and session boundaries: anti-patch reconcile-on-write with LLM-judged de-duplication, a forced PROGRESS.md handoff, a live PLAN.md anchor with an enforced directive ledger, FTS5 search, MCP tools and a GUI. Pure stdlib Python, zero runtime dependencies.
MIT ★ 10 -
lossless-claudegithub.com/offendingcommit/lossless-claudeupdated this year
Cross-runtime memory bridge between Claude Code and OpenClaw via the lossless-claw LCM SQLite database. MCP recall tools, PreCompact capture hook, SessionStart bootstrap, lcm-recall skill.
Apache-2.0 ★ 1 -
Stememgithub.com/JOJO-Adam/Stemem#readmeno update data
Identity-layer MCP runtime for AI agents: persistent personality state, emotion-driven tone drift, compression-immune memory. Local-first, zero egress, zero dependencies.
smithery -
Offline memory layer for Claude Code that indexes local session history and enables fast hybrid search (keyword + semantic) via a secure, fully offline Docker runtime.
MIT -
@noelclaw/mcpgithub.com/noelclaw/mcpupdated this year
The runtime layer for Agentic AI. 121 MCP tools: persistent memory, autonomous agents, vault storage, scheduled workflows, DeFi on Base (base_mcp_*), Robinhood Chain tokenized stocks and arbitrary crypto with V3/V4 smart routing (rh_*), live market data,
npm MIT -
mcp-dotnet-diagnosticsgithub.com/aayushmdesai/mcp-dotnet-diagnosticsupdated this year
MCP server for .NET runtime diagnostics — memory, GC, threads, and more
nuget MIT ★ 2 -
Local browser intelligence runtime with semantic memory, verified workflows, multi-surface understanding, CLI, TUI, MCP, and Rust APIs
cargo MIT -
Local-first Jarvis assistant, experimental ARDA agent runtime, and shared memory service.
★ 1