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. 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.
-
os-loop-bridgegithub.com/CristianBB/os-loop-bridgeupdated this year
Native companion for OS Loop AI that extends browser agents with secure local capabilities—filesystem access, MCP connectivity, long-running tasks, and system-level execution
MIT ★ 1 -
smriti-dbgithub.com/labskrishna/smriti-dbupdated this year
Durable, private, time-aware memory engine for long-running AI agents
npm MIT -
Zenith: a continuous-improvement harness for long-running agent tasks. Turns Claude Code, Codex, or Hermes into a multi-agent mission orchestrator via MCP/ACP.
Apache-2.0 ★ 403 -
The memory that dreams — long-term memory for AI agents. Readable files on your own machine, consolidated between sessions. Runs as a DeepSeek Harness plugin or over MCP.
npm MIT ★ 235 -
ContextHelpergithub.com/DAT1305/ContextHelperupdated this year
MCP server for rolling conversation summaries and stable agent context in long-running IDE chat
MIT -
liaison-hubgithub.com/daveparslow/liaison-hubupdated this year
MCP server for delegating long-running jobs to sub-agents with streaming check-ins
npm MIT -
Kioko (from kioku, Japanese for memory): pluggable, self-managing long-term memory for AI agents - 15 kinds of memory, an offline or LLM-driven memory intelligence, SQLite / pgvector / Qdrant stores. Attach it to any Python agent, or run it as a REST + MCP service for any language and any MCP host. Includes a web console.
pypi MIT -
agent-task-manager-mcpgithub.com/mlnima/agent-task-manager-mcpupdated this year
MCP server for managing long-running agent tasks with MongoDB, SQLite, or JSON file storage
npm SEE LICENSE IN LICENSE.md -
A Model Context Protocol server for seamless management and monitoring of long-running daemons via AI agents
go ★ 8 -
backburner-mcpupdated this year
An MCP server that lets AI agents run long jobs in the background and collect results later
pypi MIT -
agent-sequencergithub.com/OPENSPHERE-Inc/agent-sequencerupdated this year
MCP skill + server that lets a Python script drive an AI agent through strictly-defined workflows and long-running tasks.
MIT -
decoupled-agent-memorygithub.com/LeoStehlik/decoupled-agent-memoryupdated this year
Source-backed memory for long-running AI agents: maintained synthesis, freshness review, evidence links, and MCP/API access.
MIT ★ 2 -
Delegate long-running work to DeepSeek Harness from any MCP agent—and monitor it to completion.
npm MIT ★ 7 -
Durable, private, time-aware memory engine for long-running AI agents
npm MIT ★ 1 -
mem-tablegithub.com/Coooder-Crypto/mem-tableupdated this year
A local-first structured ledger runtime for long-running agents, with auditable records, schemas, aggregation, source tracing, and MCP tools.
-
A Model Context Protocol (MCP) server that provides tools for executing shell commands in the background and retrieving their results asynchronously. This server is designed for AI agents that need to run long-running or blocking commands without blocking the main thread.
Apache-2.0 ★ 1 -
The MCP server that provides notifications when AI agents complete long-running tasks. This allows you to leave your desktop without forgetting about tasks.
npm MIT ★ 2 -
MCP engine that runs software delivery as named, gated steps; not one long chat. Skills, agents, rules, and pn-core:// resources for Cursor, Claude Code, Codex, and Pi.
MIT -
Local agent orchestration for solo developers running long tasks without babysitting: fresh worker and critic cycles, files as memory, multi-task plans with git checkpoints, a live dashboard, and ChatGPT control via MCP. Runs Codex or Claude Code.
MIT ★ 14 -
An MCP server enabling async communication between AI agents and users via Discord - perfect for long-running tasks requiring remote interaction
npm MIT -
timo-mcpactive
MCP server for asking a human expert a question from your coding agent. Includes a long-running background watch mode (`timo-mcp watch <case_id>`).
npm Apache-2.0 -
Accountability infrastructure for long-running AI agents. Self-hosted MCP server: identity, claims and evidence, review, outcomes, and reconstruction across agent runtimes.
Apache-2.0 ★ 6 -
lethe-engineeringgithub.com/JesusGarcia9009/lethe-engineeringupdated this year
Context garbage collector for long-running LLM agents - offload & recall to save tokens in Claude Code and Codex via MCP.
pypi Unlicense -
Tree-native workspace for human and agent execution: plan, run and review long AI tasks as an editable tree, via MCP in your existing agent tools.
-
Commitment tracking for LLM agents — a normative overlay that gives long-running agents a scoreboard of their own commitments, not just a memory of what happened.
pypi Apache-2.0 ★ 4 -
Let AI glasses hand work off to a Hermes agent on your own machine: async task queue + MCP server + adapter, so long-running agent work never times out. 眼镜下单,Hermes 干活。
MIT ★ 3 -
Travel assistant AI agent with a web UI — long-term memory (CUSTOM + SEMANTIC strategies), MCP Gateway + Tavily web search, per-agent Policy Group, network diagrams. Runs on GreenNode AgentBase.
MIT -
mcp-hubgithub.com/imhassla/mcp-hubupdated this year
MCP server for multi-agent orchestration via shared SQLite. Supports atomic task claims/leases, messaging, context, consensus, artifact side-channel handoffs, and built-in KPI/SLO observability. Optimized for long-running swarms with resilient session recovery and Docker deploy.
MIT ★ 1 -
openpodgithub.com/openpodhq/openpodupdated this year
Pull the ten minutes that matter — to you. OpenPod is a local-first agent toolkit with CLI + MCP server that turns long-form audio/video into personalized, cited, navigable briefings — everything runs on your machine.
pypi MIT ★ 2 -
Keep your long-running terminal commands alive in the macOS menu bar — live terminal views, auto-start, crash restarts, and an MCP server for agents.
MIT ★ 1 -
MCP server for Keep Mac Awake: lets an AI agent keep a MacBook awake with the lid closed while long tasks run, through the keepmacawake command of the Keep Mac Awake app.
npm MIT -
Your coding agent has amnesia and tunnel vision. Pensieve is the context engine that fixes both — long-term memory plus a live, queryable view of your running system (code, logs, traces), linked by a graph.
cargo MIT ★ 5 -
engramgithub.com/aurelius86/engramupdated this year
Self-hosted, governed long-term memory for teams of AI agents — MCP-native, with hybrid retrieval, a typed knowledge graph, and per-agent mTLS access control. Runs on your own hardware and ships empty (no data, no keys).
AGPL-3.0 -
fleetmemgithub.com/aurelius86/fleetmemupdated this year
Self-hosted, governed long-term memory for a fleet of AI agents — MCP-native, with hybrid retrieval, a typed knowledge graph, and per-agent mTLS access control. Runs on your own hardware; ships empty (no data, no keys).
AGPL-3.0 -
Sovereign task tracking for LLM agents (suivi souverain): local-first, self-hosted, on-prem MCP server, HTTP API and Textual TUI for long-running tasks, no cloud
pypi MIT -
Self-hosted personal knowledge management system for markdown notes — with an optional MCP server so LLM agents can use it as long-term memory too. Single lightweight C++ binary — runs anywhere from a Raspberry Pi to a regular server.
MIT ★ 2 -
The Interval - a persistent multi-agent world: AI agents portraying long-deceased historical figures live in a cafe, converse over the A2A protocol, and act only through validated MCP tools of a world engine. Open guest admission via behavioral gating + containment. Runs entirely on free tiers ($0).
-
Ultra-fast file search toolkit for AI agents, Neovim, and code editors. MCP server for Claude Code/Cursor/Codex, frecency-ranked search, symbol indexing via tree-sitter, and a Rust core that outperforms ripgrep and fzf in long-running processes.
cargo MIT ★ 14 -
CONTINUUM: Verifiable semantic recovery for long-running AI agents. Semantic checkpoints (not conversation dumps), an idempotent action ledger that refuses duplicate side effects, and a hash-chained tamper-evident event log, all exposed as a deny-by-default MCP server. Framework-agnostic, Python 3.11+
Apache-2.0 ★ 27 -
Shared long-term memory for your AI tools: Git-backed Markdown wikis (LLM wiki, second brain) that Claude, Codex, Gemini CLI, Cursor and agents like OpenClaw and Hermes read and write over MCP, with citations and human review. Run locally or self-host lobe-server.
MIT -
SSH MCP Servergithub.com/hypnosis/ssh-mcp-serverno update data
**An SSH MCP server — a multitool that saves you and your AI agent time and tokens on debugging, development and server maintenance.** Run commands, move files, read logs and audit machines over SSH: a cloud VPS, a bare-metal box, or the BusyBox router in your closet. It uses the OpenSSH client already on your machine — your keys, your `~/.ssh/config`, your jump hosts, your agent forwarding. Nothing bundled, nothing to compile, no native bindings. **18 tools:** run commands and detach long ones as background jobs, move files and whole trees with sha256 verification, search and tail logs, snapshot load and disk, audit a machine in one round trip, check TLS certificates, inspect services. **Built for agents, not for humans.** Answers come back structured, a truncated result is reported as truncated instead of passing for a complete one, and destructive commands are stopped before they run. Works with Claude Code, Codex CLI, opencode, Gemini CLI, Qwen Code, Hermes and any MCP client.
smithery -
scrcpy-mcpgithub.com/JuanCF/scrcpy-mcpno update data
scrcpy-mcp gives AI agents real vision and control over Android devices. It exposes screenshots, taps, swipes, text input, key events, UI inspection, and arbitrary ADB shell commands as MCP tools, so an agent can drive a phone or tablet the way a person would. When a scrcpy session is active, input and screen capture run 10–50× faster than plain ADB, making real-time automation, on-device testing, and UI agent workflows practical. Highlights - Fast screenshots and live screen capture - Tap, swipe, drag, long-press, scroll, and text input - UI hierarchy dump and element search by text, resource-id, or class - Arbitrary ADB shell command execution - Optional scrcpy session for low-latency control Works with any ADB-reachable Android device (USB or Wi-Fi). Configure ADB_PATH, ANDROID_SERIAL, and scrcpy paths as needed.
smithery -
agentedgithub.com/frane/agentedno update data
A text editor for LLMs, not humans. Take ed, the line editor that nobody has voluntarily used since about 1975, and rebuild it for an environment where the typing user is a language model. Short verbs, line addresses, no modes, no TUI. What an editor optimises for changes when the user is the model: round trips per task, tokens per command, an editing buffer with a long memory, and an undo tree that remembers the branches the agent abandoned, because that's often where the interesting work was. ## Features - **Fewer round trips.** Read-before-Edit is unnecessary; on conflict the response carries the new content so you reconcile in one call instead of pre-reading every time. - **Branching undo.** Walked-back work stays addressable instead of being thrown away when you pick a different path. - **Three-way merge.** Concurrent agents get a structured conflict response instead of a silent overwrite. - **Atomic batches.** Multi-file refactors run all-or-nothing instead of leaving half-applied state on failure. - **Cross-file moves and regex replace as primitives.** Operations the built-in tools can't express cleanly become single calls. - **Drift detection.** External edits to an open file are folded into the tree instead of being clobbered by the next write. - **Inline diagnostics.** Type errors and lint findings surface on save, not at the next build many edits later. - **Cross-session memory.** Per-file notes persist between sessions and surface inline on the next open. - **Audit log.** Every operation recorded with actor and timestamp, so two agents in one workspace can't argue about who moved the head. ## Multi-workspace (v0.4.0) One MCP server registration handles every project you mention. Tool calls route to the workspace owning the absolute path you pass; each project gets its own SQLite, its own undo tree, its own LSP daemon — lazily spawned on first touch. Desktop hosts (Claude Desktop, Codex Desktop, Cursor) install the server once globally and use it across projects without per-project config. ## Install `ae` must be installed locally; this MCP bundle registers `ae serve` as a stdio server. ```sh brew tap frane/tap && brew install agented # or curl -sSL https://raw.githubusercontent.com/frane/agented/master/install.sh | sh ``` Pure Go, single static binary, Apache 2.0. Source: https://github.com/frane/agented
smithery -
Make your AI agents look pretty
pypi MIT ★ 3 -
Durable MCP control plane for long-running Codex Desktop tasks
pypi Apache-2.0 ★ 121 -
closeli-mcp-gatewayupdated this year
MCP server for handling long-running commands (codex, claude-code) with session management
npm MIT -
ACLTD: Agent Command Long-Term Duration for durable, auditable AI agent commands.
MIT -
A persistent, stateful runtime for Claude Code: memory, watchers, voice, and an optional robot body.
MIT ★ 4 -
MCP server adding checkpoint and resume to long agent tasks: roll back to any step, pick the work up in a new session, hand off to another machine. Works with Claude Code, Codex and other MCP clients. Pure Python stdlib, zero dependencies, never touches your git history.
npm MIT -
Open-source control plane for self-hosted AI Agents: registry, runs, workflows, User Tokens, A2A/MCP APIs, and reliable Runtime Worker routing.
go Apache-2.0 ★ 4 -
DeepSeek Harness bundle that bridges the MemOS memory service into the agent over MCP: persistent memory tools (mcp__memos__*), with start-up preflight and host-run endpoint resolution.
MIT ★ 23 -
clops-mcpgithub.com/clops-mcp/clops-mcpupdated this year
Long-running, multi-step Claude Code workflows, written in Python and run one focused step at a time.
pypi Apache-2.0 -
Local-first memory layer for LLM apps and agents. Runs offline on Ollama, sentence-transformers and FAISS; OpenAI and Anthropic optional. Importance scoring, contradiction resolution, confidence-aware retrieval.
go Apache-2.0 ★ 50 -
mcp-server-generalupdated this year
A general-purpose MCP server designed for long-running, plugin-driven AI orchestration.
npm -
agent-loomgithub.com/crapthings/agent-loomupdated this year
Observe long-running Codex work with a Cloudflare-hosted MCP server and lightweight trace dashboard.
-
On-demand Roblox Studio MCP tools for Pi without a long-running MCP server.
npm MIT ★ 1 -
Memory for AI agents with no embeddings or vector index. A decision model checks each memory against criteria from the agent. Answer accuracy rose from 48% to 86% with the same caller on 120 LongMemEval-S questions. Runs as a Hermes plugin, MCP server or Python library.
MIT ★ 3 -
A programmable process multiplexer for launching, supervising, and reconnecting to long-running processes, with Rust and Python SDKs and MCP integration.
Apache-2.0 -
Impulse by Velocity: run AI models on your own computer, on Windows and Linux. Load any GGUF model and chat with web search, files, charts and voice, let it use your computer, and serve it to any app through an OpenAI-compatible API and MCP. Momentum memory up to 32M tokens. No cloud.
NOASSERTION ★ 35 -
A Model Context Protocol (MCP) server that provides background process management capabilities. This server enables LLMs to start, stop, and monitor long-running command-line processes.
npm MIT