Fit LLM MCP Server

FitLLM is an open-source, zero-dependency engine that checks whether a local LLM fits on a GPU or Apple Silicon Mac using architecture-aware memory math. Its 3 documented tools cluster into check, what, read & search. It is available as a remote MCP endpoint.

Teams that work with apple silicon, kv cache and memory calculator and want a hosted endpoint instead of running a local process. The project is written in JavaScript.

VERIFIED ACTIVE

LAST COMMIT 2026-09-14 · ★ 8 · #245 OF 304 MAINTAINED MEMORY · VERIFIED 2026-09-18

MIT · JavaScript servers · how we verify → /methodology

01 · Install Fit LLM

before you install - you'll need

The README does not document required environment variables for a basic install.

Claude Code

claude mcp add click6067-ship-it-fitllm-engine --transport http https://fitllm.run/api/mcp

Claude Desktop / Cursor / VS Code - add to config

{
  "mcpServers": {
    "click6067-ship-it-fitllm-engine": {
      "url": "https://fitllm.run/api/mcp"
    }
  }
}

Same JSON for Cursor. For VS Code, rename the top-level key from `mcpServers` to `servers`.

Using another client? Same JSON, different key

Claude Desktop · mcpServers

Cursor · mcpServers

VS Code · servers

Windsurf · mcpServers

Zed · context_servers

Cline · mcpServers

Roo Code · mcpServers

Continue · mcpServers

LibreChat · mcpServers

Gemini CLI · mcpServers

Codex CLI · mcp_servers

Full setup guides: every client.

02 · Evidence

https://fitllm.run/api/mcp

transport: streamable-http

endpoint alive - responded to MCP initialize · probed 2026-09-18

Security posture

What to check before giving this server access to your agent - from the registry, GitHub, and our own probes. We don't score safety; we show what's verifiable.

runs as remote - your agent's requests go to fitllm.run

endpoint auth accepted our unauthenticated MCP initialize - no credentials needed to connect

license MIT - declared in the repository

registry vendor namespace run.fitllm - domain-verified with the official MCP registry

03 · What Fit LLM can do

Prose above is summarized from the project's README and registry record - no invented capabilities.

What you can build

With this server connected, an agent can check one model against a GPU, multi-GPU rig, or Mac and return the verdict, memory breakdown, and a fix when it does not fit, and list the built-in model and hardware names accepted by the fit checker.

The 3 tools it gives your agent

Extracted from the project's README - what FitLLM lets an agent do.

check_llm_fit: - check one model against a GPU, multi-GPU rig, or Mac and return the verdict, memory breakdown, and a fix when it does not fit.
what_fits_on_hardware: - rank the supported local models that fit the given GPU, multi-GPU rig, or Mac.
list_supported: - list the built-in model and hardware names accepted by the fit checker.

Latest releases

v2.16.0 · 2026-09-14

Catalog refresh for the models that landed since 2026-09-05, plus two fixes to the discovery loop that nearly made us miss them. · Added · | Model | Params | How it was checked | · |---|---|---| · | MiniCPM5-2B |…

v2.15.0 · 2026-09-05

What changed · Fail-closed Gemma 4 PLE parsing. Partial or malformed profiles, invalid vocab_size, and model-name-only parameter guesses can no longer certify the PLE GPU-weight deduction. · Truthful residency scope.…

v2.14.1 · 2026-09-05

What changed · Aligns the gpt-oss-20b and gpt-oss-120b catalog totals with FitLLM’s existing Hugging Face safetensors evidence path: 20.9B and 116.8B logical parameters. · Preserves the fail-closed MXFP4 behavior: when…

04 · Who maintains Fit LLM

FitLLM is maintained by click6067-ship-it. It's the only MCP server we track from this author; the repo dates to May 2026.

05 · Facts

repository: github.com/click6067-ship-it/fitllm-engine
website: https://fitllm.run/
category: memory - ranked #245 of 304 actively-maintained memory servers as of 2026-09-18.
release cadence: 10+ releases in the last 90 days (latest 2026-09-14)
registry: run.fitllm/fitllm (active, first published 2026-07-09 · 2 versions)

06 · Fit LLM FAQ

Is Fit LLM still maintained?

Yes - as of 2026-09-18, its last commit was 2026-09-14 and it shipped 10+ releases in the last 90 days. We re-verify nightly.

What can Fit LLM do?

How do I install Fit LLM?

Run `claude mcp add click6067-ship-it-fitllm-engin --transport http https://fitllm.run/api/mcp`. The README does not document required environment variables for a basic install. You can also paste the ready-made client config above.

Does Fit LLM require authentication?

No - the endpoint accepted our unauthenticated MCP initialize when probed on 2026-09-18; you can connect without credentials.

07 · Alternatives to Fit LLM

Alternatives to Fit LLM

Maintained memory servers if Fit LLM isn't the fit.

Pairs well with

Servers that cover what Fit LLM doesn't - only shown when the pairing reason fits the companion.

More memory MCP servers · Cpersona · Growmos · Agent Coherence - Stale Write Guard (FS) · Openfda MCP Server · KLYPIX Project Brain

More JavaScript MCP servers · Sats4AI - Bitcoin-Powered AI Tools · Procm · Dataverse · Dex Paprika · Coinversaa MCP Server · see all