Luftpause

Name: Luftpause
Author: luftpause

by luftpauseversion 0.1.4i★ –iRuns on your computer

Registry nameio.github.vmedvedovskiy/luftpause

Install

Overview

Description from the maker (original English)

async/await for LLM agents: spawn and await sub-agents scheduled on your own GPUs

PopNx MCP’s three pillars

These three are never merged into one score: opinions, measurements and popularity are different things.

User voiceiNo reviews yetReviews open together with accounts in a later step
Our basic checksi0 of 1 passedChecked from public information; code not scanned yet
UsageiNot measured yet★ –

What you can do with it

We have not written a plain-language summary for this one yet. Below are the facts we read from the registry.

Connection: Runs on your computer
Transport: Streamable HTTP
Category: Search & AI memory
You need to fill in: Runtime__Model, LUFTPAUSE_TOKEN

Install

Pick the AI app you use, then copy the settings.

Before you install

Basic checks: 0 of 1 passed
Runs on your computer with your user’s permissions (files, network, apps).
You fill in yourself: Runtime__Model, DataPlane__Url, LUFTPAUSE_TOKEN

Values marked secret are keys to your account. Never share them or paste them into a chat.

You need

DockerDocker installed

Which app do you use?

ClaudeClaude CodeCursorVS CodeChatGPT

Pick your app above. We remember it, so next time the button says “Install in …”.

Open Claude Desktop → Settings → Developer → Edit Config, and paste this into claude_desktop_config.json

{
  "mcpServers": {
    "luftpause": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "-e",
        "Runtime__Model",
        "-e",
        "DataPlane__Url",
        "-e",
        "LUFTPAUSE_TOKEN",
        "ghcr.io/luftpause/runtime:v0.1.4"
      ],
      "env": {
        "Runtime__Model": "your-value",
        "DataPlane__Url": "your-value",
        "LUFTPAUSE_TOKEN": "your-value"
      }
    }
  }
}

Run this in your terminal

claude mcp add luftpause --env Runtime__Model=your-value --env DataPlane__Url=your-value --env LUFTPAUSE_TOKEN=your-value -- docker run -i --rm -e Runtime__Model -e DataPlane__Url -e LUFTPAUSE_TOKEN ghcr.io/luftpause/runtime:v0.1.4

Fill in the empty values before running it.

Open in Cursor

Or copy the settings yourself

Paste into .cursor/mcp.json in your project, or ~/.cursor/mcp.json

{
  "mcpServers": {
    "luftpause": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "-e",
        "Runtime__Model",
        "-e",
        "DataPlane__Url",
        "-e",
        "LUFTPAUSE_TOKEN",
        "ghcr.io/luftpause/runtime:v0.1.4"
      ],
      "env": {
        "Runtime__Model": "your-value",
        "DataPlane__Url": "your-value",
        "LUFTPAUSE_TOKEN": "your-value"
      }
    }
  }
}

Your browser asks to open the app, then the app shows the settings and asks you to confirm. Key values are left empty for you to fill in.

Open in VS Code

Or copy the settings yourself

Paste into .vscode/mcp.json in your project

{
  "servers": {
    "luftpause": {
      "type": "stdio",
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "-e",
        "Runtime__Model",
        "-e",
        "DataPlane__Url",
        "-e",
        "LUFTPAUSE_TOKEN",
        "ghcr.io/luftpause/runtime:v0.1.4"
      ],
      "env": {
        "Runtime__Model": "your-value",
        "DataPlane__Url": "your-value",
        "LUFTPAUSE_TOKEN": "your-value"
      }
    }
  }
}

Your browser asks to open the app, then the app shows the settings and asks you to confirm. Key values are left empty for you to fill in.

This one runs on your computer, and ChatGPT can only connect to servers on the internet. Pick another app.

Your choice stays on this device in a cookie; no account needed.

Values you fill in

Runtime__Model required env
Model the sub-agents run on, as served by an engine behind the luftpause dataplane
DataPlane__Url env
URL of the luftpause dataplane that places each turn on the engine pool
LUFTPAUSE_TOKEN secret env
Bearer token required on the MCP endpoint and sent to the dataplane; unset leaves both open

These settings are generated from the maker’s registry entry. If the README says otherwise, follow the README.

What it can do

Runs on your computer; the tool list appears in your AI app after you install it.

Basic checksi

We check information anyone can see; we have not read the code. Checks that do not apply are left out of the count. · Checked Nov 7, 2026

! Caution
Maker matches the repo
The publisher (vmedvedovskiy) is not the repo owner (luftpause); someone else may have listed it
… Not checked yet
Still maintained
We have not read this repo from GitHub yet
… Not checked yet
Has a licence
We have not read this repo from GitHub yet
– Not applicable
Endpoint responds
This one runs on your computer; there is no internet endpoint
– Not applicable
Sign-in with OAuth
Runs on your computer and uses local settings instead
… Not checked yet
No hidden instructions in tool descriptions
Runs on your computer; we can only read tool lists from internet endpoints

Compared with others in Search & AI memoryi

More stars than 0% of the other 7257 servers in this category
Published to the registry more recently than 73% of the other 7257 servers in this category
Basic checks equal to or better than 31% of the other 7257 servers in this category

Compared with every active server in this category in the registry. Stars count only when the publisher owns the repo.

Common questions

Is Luftpause safe?

It passed 0 of 1 applicable basic checks. We have not read the code. Before adding a key or token, see what access the server asks for and use a key with only the permissions you need.

What keys or codes does it need?

You need to fill in: Runtime__Model, LUFTPAUSE_TOKEN

Does it run on my computer or in the cloud?

It runs on your computer. Your AI app starts it when needed, and data does not pass through the maker’s servers (unless it calls an online service itself).

Is it still maintained?

The GitHub repo last changed –, and the latest version (0.1.4) reached the registry 14 days ago.

Which apps can use it?

Apps that support MCP, such as Claude Desktop, Claude Code, Cursor and VS Code. You also need Docker on your computer.

What can I use instead?

In the same category: Context7, Codebase Memory, Exa

Alternatives

Servers in the same category from other makers, sorted by stars.

Reading a card:★ GitHub starsiThree pillars: reviews · checks · usageiUpdated = latest version published to the registryi

Context7

by upstash

Feeds AI the latest docs for libraries and frameworks before it answers, so code it writes does not use outdated, removed functions.

Computer or cloud★ 64,224MITUpdated today

Install

Codebase Memory

by DeusData

Codebase knowledge graph for AI agents — 162 languages, sub-ms queries, 99% fewer tokens.

Runs on your computer★ 48,148MITUpdated 20 days ago

Install

Exa

by exa-labs

Connect AI agents to Exa for web search, content fetching, and multi-step research.

Runs in the cloud★ 5,213MITUpdated last month

Install

README from the maker

Open the README

Copied from the maker’s repo as written. PopNx MCP has not reviewed or endorsed it. HTML, images and scripts are removed. Links lead off this site.

luftpause

The runtime layer under your agent harness. Every tool call is an `await`; the engine forgets the run, luftpause keeps it.

Tool calls split an agent into turns. To an inference engine each turn is a new request from nobody: when the tool returns, the continuation queues behind every batch request that arrived while it waited, and its KV cache may be gone. luftpause sits between the harness and the engines and keeps the run:

Continuations go first. A run paused on a tool call resumes ahead of batch work; batch turns leave room for the interactive runs that will come back.
A run stays where its prefix is. While a run is generating or paused on a tool, its turns go only to the replica that holds its KV cache; new turns are admitted by the engine's own queue.
One caller cannot starve another. Held turns drain fairly across callers.
Any harness, two doors. Point base_url at it (OpenAI-compatible, your own loop), or connect over MCP and get spawn / await sub-agents that survive a restart.

Measured on one 8 GB laptop GPU, engines saturated (48 batch runs flooding, 4 interactive runs beside them):

	Throughput	Latency	What it costs
Agents with their own loop, through luftpause vs straight to the SGLang router	2.0× interactive (12.7 vs 6.4 runs/min)	interactive p50 19 s instead of 36 s	batch −9 %
Sub-agents over MCP, luftpause vs the runtime straight to the router	2.7× interactive (8.0 vs 3.0 runs/min)	interactive p50 27 s instead of 58 s	batch unchanged
Fan-out, 16 sub-agents, vs one at a time	5.3× (all 16 in 23 s instead of 120 s)	per sub-agent p50 15–22 s, the same as through the router	nothing: the router alone takes 21.5 s
Claude Code as the client, six Wikipedia look-ups	2.9× (12.5 s instead of 36 s back to back, unprompted)	70B sub-agents: all six in one round, 9.3 s	0 failed tool calls, 0 re-spawns

Try it

docker compose up -d                # fake engines, no GPU needed

claude mcp add --transport http --scope local luftpause http://localhost:7000/mcp

docker compose up -d pulls the published images; add --build to build them from source instead (first build takes a few minutes). It pulls latest; pin a release with LUFTPAUSE_VERSION=v0.1.4 docker compose up -d. The runtime is also listed in the official MCP Registry as io.github.vmedvedovskiy/luftpause.

Claude Code now has spawn, await, run, status, cancel, approve, tools and decide. The fake engines answer with filler, and the sub-agents' only tools are fake-mcp's lookup and deep_search, which wait a random second or so and return filler too — enough to see the fan-out. For real answers from an Ollama already on your machine, with Wikipedia as the sub-agents' tool instead of the fake ones:

docker compose -f compose.yaml -f compose.ollama.yaml up -d scheduler dataplane runtime

Wikipedia's API rate-limits anonymous callers, so a tool call can take tens of seconds while wikipedia-mcp retries. Llama 3.2 1B sometimes passes a tool's schema instead of its arguments; the runtime rejects the call and the sub-agent may answer without it, so ask again or pull a larger model.

Ollama needs the model pulled (ollama pull llama3.2:1b, or set OLLAMA_MODEL) and must listen beyond loopback (OLLAMA_HOST=0.0.0.0) unless Docker Desktop already reaches it (it does on Windows and macOS with the default loopback binding). Ollama serves one request at a time unless OLLAMA_NUM_PARALLEL is raised, so sub-agents queue behind each other there; luftpause sees no queue metric from it and does not cap it. SGLang on your GPU (compose.testbed.yaml) or any OpenAI-compatible endpoint (compose.runpod.yaml shows a hosted vLLM pod) work the same way.

How it differs

Routers place requests on a GPU pool; MCP servers delegate a sub-task to a local model. Neither knows what a run is. luftpause does: it sees every turn and every tool call of every run, through either door, and schedules them on one shared pool with three facts an engine never has: that a paused run will come back, what class of work it is, and who is asking.

The gains come from admission, not from routing. Routing alone ties SGLang's cache-aware router (below); with the engines idle, every arm ties. The difference appears when the engines are saturated, which is when it matters.

Two ways in

OpenAI-compatible proxy — an agent that keeps its own loop points its base_url at luftpause. It sends x-luftpause-run-id per run and, optionally, x-luftpause-class and x-luftpause-caller; every tool call in its loop is then a scheduled await.
MCP runtime — an agent such as Claude Code connects over MCP and gets spawn / await. luftpause runs each sub-agent's loop through the same proxy: LLM turns on the pool, tool calls to upstream MCP servers, the MCP session as the caller, and a journal so the runs outlive the process.

examples/ has three own-loop agents that set all three headers: the openai package, the OpenAI Agents SDK and LangGraph.

Connect from

A harness that cannot set a header per run still works through the proxy: the run is recognised by the tool_call_ids it echoes back, and a fixed x-luftpause-caller is enough for fairness.

Tool	MCP door	Proxy door
Claude Code	`claude mcp add --transport http luftpause http://localhost:7000/mcp`	n/a: speaks the Anthropic Messages API, not Chat Completions
Codex CLI	`~/.codex/config.toml`: `[mcp_servers.luftpause]` `url = "http://localhost:7000/mcp"`	n/a: `wire_api` accepts only `responses`; the proxy serves Chat Completions
Gemini CLI	`gemini mcp add --transport http luftpause http://localhost:7000/mcp`, or `.gemini/settings.json`: `{"mcpServers": {"luftpause": {"httpUrl": "http://localhost:7000/mcp"}}}`	n/a: speaks the Gemini API (unverified)
Goose	`goose session --with-streamable-http-extension "http://localhost:7000/mcp"`, or `goose configure` → Add Extension → Remote Extension (Streamable HTTP)	`GOOSE_PROVIDER=openai OPENAI_HOST=http://localhost:8080 OPENAI_API_KEY=x OPENAI_CUSTOM_HEADERS="x-luftpause-caller=goose"`
Cursor	`~/.cursor/mcp.json`: `{"mcpServers": {"luftpause": {"url": "http://localhost:7000/mcp"}}}`	n/a: "Override OpenAI Base URL" is called from Cursor's servers, which cannot reach localhost (unverified)
Zed	`settings.json`: `"context_servers": {"luftpause": {"url": "http://localhost:7000/mcp"}}`	`settings.json`: `"language_models": {"openai_compatible": {"luftpause": {"api_url": "http://localhost:8080/v1", "available_models": [{"name": " ", "max_tokens": 128000}], "custom_headers": {"x-luftpause-caller": "zed"}}}}`
OpenHands	Settings → MCP Servers → add `http://localhost:7000/mcp`	Settings → LLM → Advanced: Custom Model `openai/` , Base URL `http://localhost:8080/v1`, any API key; no header setting
Cline / Roo Code	Cline `cline_mcp_settings.json`: `{"mcpServers": {"luftpause": {"type": "streamableHttp", "url": "http://localhost:7000/mcp"}}}`; Roo `.roo/mcp.json`: same with `"type": "streamable-http"`	Provider "OpenAI Compatible", Base URL `http://localhost:8080/v1`, any API key; `x-luftpause-caller` under Custom Headers (unverified)

Architecture

%%{init: {"themeVariables": {"fontFamily": "-apple-system, Segoe UI, Helvetica, Arial, sans-serif"}, "flowchart": {"wrappingWidth": 260}}}%%

flowchart LR mcp["Claude Code or any MCP client"] own["Agents with their own loop"]

subgraph lp["luftpause"] runtime[" runtime · .NET spawn · await · journal sub-agent loops"] dp[" dataplane · Rust OpenAI-compatible proxy"] sched[" scheduler · .NET decide(state, event)"] end

tools[("MCP tool servers")]

subgraph pool["Engine pool"] direction TB engines["SGLang · vLLM"] router["SGLang router"] hosted["Hosted OpenAI- compatible APIs"] end

mcp -- MCP --> runtime own -- "HTTP / SSE" --> dp runtime -- "tool calls" --> tools runtime -- "LLM turns" --> dp dp -- "turns" --> pool sched <-- "Admit · Report · run ended" --> dp sched -. "health · load" .-> pool

runtime (runtime/) — MCP server. Each spawned run is a .NET Task running the agent loop; tool calls go to the configured MCP servers. Every run is journaled to runs.jsonl and resumed on restart; a finished run is reported to the dataplane (DELETE /v1/runs/{id}) so the scheduler forgets it at once.
dataplane (dataplane/crates/dataplane) — tokio/axum proxy. Resolves the run (header or tool-call ids), reads the class and caller headers, asks the scheduler where the turn goes, streams the engine's response through unchanged and reports first token, usage and finish reason.
scheduler (scheduler/) — the decision core is pure functions over immutable state (Luftpause.Scheduling); the gRPC host executes the actions.
test kit — fake-engine (OpenAI + /classify with simulated latency), fake-mcp (a lookup and a slow deep_search tool), tools/loadgen (agents, sub-agent fan-out over MCP, closed-loop throughput runs).

What it does

Capability	How
`spawn` / `await` over MCP	`spawn(task, model?, serviceClass?, system?, maxToolCalls?, tools?, maxTokens?, parallelToolCalls?, parent?, budgetTokens?, budgetToolCalls?, budgetSeconds?, approveTools?)` returns a handle at once; `await(handles, all\	any, timeout) `collects results with` toolCalls `/` failedToolCalls`; also` tools`,` status`,` cancel`, and` decide(input, labels?, model?)`: without` labels `the single-pass model's labels, with 2 to 20` labels `one` max_tokens: 1 `turn on the generative model scored by first-token logprobs, returning` label`,` probabilities `and` unscored`
Durable runs	every run is journaled (`runs.jsonl` under `Runtime:StateDir`): spawned, each tool round's messages, finished; on restart the runtime replays the journal, answers `await` for finished runs and resumes unfinished ones from their last complete tool round
Run tree	`spawn(parent)` makes a run a child; `status` lists `parent` and `children`; `cancel` of a parent cancels its running descendants; the runtime's `GET /status` returns the tree with each run's own work and its subtree's totals (turns, tool calls, failed tool calls, prompt and completion tokens, turn and tool seconds)
Budgets	`spawn(budgetTokens?, budgetToolCalls?, budgetSeconds?)` caps a run's whole subtree (it and every run spawned under it, later ones included) on prompt plus completion tokens, tool calls and wall-clock seconds; checked on every turn and tool round anywhere below a budgeted run, and by a timer for seconds so an idle subtree still trips; past a limit the subtree is cancelled with `budget exceeded: /` , and `status` / `await` show `budget` with `limits`, `used` and `exceeded`

| Tool approval | a sub-agent's tool call waits for the harness user when its server marks the tool destructiveHint: true (and not readOnlyHint: true) or spawn(approveTools) names it; the run shows status AwaitingApproval with pendingApproval {tool, arguments}. A client that advertises elicitation is asked while its await or run covers the run, with a one-field form (allow): on a session as an elicitation request, on a sessionless 2026-07-28 request as an input-required result. Other clients answer with approve(handle, allow). A declined call reaches the sub-agent as Error: the user declined this tool call. Decisions are journaled: one answered just before a restart is not asked again, an unanswered one is. Clients: Claude Code 2.1.218 in a terminal shows the prompt during await

Only the beginning is shown; read the rest in the repo

Luftpause

Overview

PopNx MCP’s three pillars

What you can do with it

Install

Before you install

Values you fill in

What it can do

Basic checksi

Maker matches the repo

Still maintained

Has a licence

Endpoint responds

Sign-in with OAuth

No hidden instructions in tool descriptions

Compared with others in Search & AI memoryi

Common questions

Alternatives

Context7

Codebase Memory

Exa

README from the maker

luftpause

Try it

How it differs

Two ways in

Connect from

Architecture

What it does