Skip to content
MCP server

Neurarch

Size GPU training jobs and lint PyTorch models inside your coding agent, offline

Runs locallyStatus and history
Install
Pricing
Free
Platforms
CLI, self-hosted
Runs
Locally
Version
0.21.0
Listing updated
2 days ago

About Neurarch

Neurarch is a free MCP server for ai models and llm tooling. It runs locally on your machine and is started by your AI client.

The pitch is to know what a job needs before it takes a GPU reservation. Given a model, batch and card, the sizing tool reports peak memory, the smallest card that holds it, the largest batch per card, a two-card split when one is not enough and the cost of a run. After a CUDA out-of-memory error, a diagnosis tool reads the message against the same estimate and names the likely cause with the least invasive fix first.

It reads a PyTorch .py file, a model.neurarch.json saved from the Neurarch app, a graph from the neurarch-trace package, or a HuggingFace config. Source files give exact layers and parameter counts but unknown tensor shapes. The README says memory is a calibrated estimate, with median error 7% above 1 GB on held-out architectures, so treat it as a fit check.

A Claude Code plugin adds a hook that sizes a python train.py launch before it runs and blocks one that will not fit. The README requires Node.js 20 or newer, and the site says no API key is needed.

Install

Run in a terminal

claude mcp add neurarch-mcp -- npx -y neurarch-mcp

Add to claude_desktop_config.json

{
  "mcpServers": {
    "neurarch-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "neurarch-mcp"
      ]
    }
  }
}
Install in CursorOpens Cursor, which asks you to confirm. Or add it by hand:

Add to ~/.cursor/mcp.json

{
  "mcpServers": {
    "neurarch-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "neurarch-mcp"
      ]
    }
  }
}
Install in VS CodeOpens VS Code, which asks you to confirm. Or add it by hand:

Add to .vscode/mcp.json

{
  "servers": {
    "neurarch-mcp": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "neurarch-mcp"
      ]
    }
  }
}

Add to ~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "neurarch-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "neurarch-mcp"
      ]
    }
  }
}

Add to your client's MCP settings (most use this shape)

{
  "mcpServers": {
    "neurarch-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "neurarch-mcp"
      ]
    }
  }
}

Tools this server provides (7)

Taken from the project documentation; the running server may differ.

size_training_job
Estimates peak memory, smallest card, largest batch and run cost.
diagnose_oom
Reads a CUDA out-of-memory message and names the cause.
lint_model
Structural checks on a model.
check_design
Design checks run by the CI bot.
suggest_fix
Turns lint findings into unified diffs.
history
What the same structure measured the last time it ran.
load_hf_model
Builds a graph from a HuggingFace config.

What it does

Training job sizing
Estimates peak memory, smallest fitting card, largest batch and run cost.
OOM diagnosis
Reads a CUDA out-of-memory message and names the cause with a fix.
Model linting
Runs structural checks on a model before training.
Fix suggestions
Turns lint findings into diffs labelled exact or proposal.
Hugging Face import
Builds a graph from a Hub config.json without downloading weights.
Launch hook
A Claude Code hook blocks training launches that will not fit.

Good for

  • Check whether a model fits a T4 or L40S before launching
  • Find the right card count for a large model
  • Diagnose a CUDA out-of-memory error
  • Find which layers hold most of the parameter budget
  • Block an agent-launched job that will not fit

Pricing

Free

Runs offline with no API key

Prices change. Check neurarch.com before you buy.

Neurarch FAQ

Does it need an API key?

No. The site says it works offline with no API key; only a named Hugging Face id opens a network connection.

How accurate is the memory figure?

It is a calibrated estimate, not a measurement. The README reports a median error of 7% above 1 GB on held-out architectures.

What does it need to run?

Node.js 20 or newer; clients launch it through npx.

Which inputs work best?

A .neurarch.json or neurarch-trace graph carries explicit structure. Plain .py parsing leaves tensor shapes unknown.

Is Neurarch free?

Yes. Neurarch is free to use.

How do I add Neurarch to Claude Code?

Run this command in your terminal: claude mcp add neurarch-mcp -- npx -y neurarch-mcp

What are the best alternatives to Neurarch?

The closest alternatives to Neurarch are REA, Codebase Memory, GitHub, Serena MCP: the IDE for your agent and XERJ.

Alternatives to Neurarch

  1. REA

    MCP server

    CLI and MCP server that lets coding agents reverse engineer apps and binaries

    Runs locally
    • Open source
    • 464K GitHub stars
    • Developer tools
  2. Codebase Memory

    MCP server

    Local MCP server that builds a code knowledge graph for AI coding agents

    Runs locally
    • Open source
    • 48.7K GitHub stars
    • Knowledge and memory
  3. GitHub

    MCP server

    Official MCP server connecting AI assistants to GitHub repos, issues and workflows

    Not checked yet
    • Remote
    • Open source
    • 34.5K GitHub stars
    • Developer tools
  4. MCP toolkit giving coding agents symbol-level code retrieval and editing

    Runs locally
    • Open source
    • 31.5K GitHub stars
    • Developer tools
  5. XERJ

    MCP server

    Local search engine for AI agents with cited answers over your code and docs

    Runs locally
    • Open source
    • 19.2K GitHub stars
    • Knowledge and memory
  6. Ouroboros

    MCP server

    Spec-first layer that pins requirements and verifies results of AI coding agent work

    Runs locally
    • Open source
    • 6.4K GitHub stars
    • Developer tools