Repository avatar
Search & Data Extraction
v0.1.4
active

matrix-compute-mcp

com.matrix-compute/matrix-compute-mcp

P2P GPU matrix multiply, exact cosine search, and PCA/SVD verified via Freivalds' algorithm.

Documentation

matrix-compute-mcp: Official MCP Server & High-Level Verifiable AI Primitives

PyPI version Open In Colab

matrix-compute-mcp is an official Model Context Protocol (MCP) server and high-level AI primitives package built on top of matrix-compute-sdk that provides:

  1. High-Level Verifiable AI Primitives (matrix_compute_mcp.ai): Drop-in PyTorch layers (DistributedLinear, DistributedLoRALinear), massive exact cosine similarity search (distributed_cosine_similarity), and distributed spectral PCA/SVD (distributed_pca_svd) — all verified in $O(n^2)$ via Freivalds' algorithm.
  2. Official Model Context Protocol Server (matrix-compute-mcp): Exposes decentralized GPU TFLOPS to autonomous AI agents (Claude Desktop, Cursor, Windsurf, Antigravity, LangChain, OpenClaw) over stdio (uvx matrix-compute-mcp) or remote Streamable HTTP (https://mcp.matrix-compute.net/mcp) with L402 Macaroon, NWC (nostr+walletconnect://), or zero-drain trial sandbox support.

Quick Start

1A. Local stdio MCP Client Configuration (Claude Desktop / Cursor / Antigravity)

{
  "mcpServers": {
    "matrix-compute": {
      "command": "uvx",
      "args": ["matrix-compute-mcp"],
      "env": {
        "MATRIX_NOSTR_RELAYS": "wss://relay.matrix-compute.net",
        "MATRIX_SETTLEMENT_API_URL": "https://settlement.matrix-compute.net"
      }
    }
  }
}

1B. Hosted Remote MCP Endpoint (Zero Install)

Point any Streamable HTTP MCP client directly to:

https://mcp.matrix-compute.net/mcp

2. Exposed MCP Tools

Tool NameDescription
get_network_statusQueries the discovery relay for live GPU worker count, total VRAM, TEE enclaves, and sats_per_gflop pricing.
estimate_job_costComputes exact upload/download bandwidth (MB), task count, GFLOPs, and satoshi cost estimates via BandwidthProfiler.
execute_distributed_matmulExecutes verified distributed matrix multiplication $C = A \times B$ from .npy, .npz, .safetensors, or inline tensors.
batch_similarity_searchRuns massive exact $K$-NN cosine similarity search using OUTPUT_GRID_2D zero-reduction tiling.
compute_pca_svdRuns verified top-$k$ randomized SVD or stateful remote-VRAM power-iteration PCA/SVD.

3. High-Level Verifiable AI Primitives (matrix_compute_mcp.ai)

import numpy as np
from matrix_compute_mcp.ai import (
    DistributedLinear,
    DistributedLoRALinear,
    distributed_cosine_similarity,
    distributed_pca_svd,
)

# 1. Exact K-NN Cosine Similarity with OUTPUT_GRID_2D zero download redundancy
scores, indices = distributed_cosine_similarity(queries, corpus_embeddings, top_k=10)

# 2. Drop-in PyTorch Linear & LoRA Linear layers with verified forward & backward passes
layer = DistributedLinear(in_features=4096, out_features=4096)
lora_layer = DistributedLoRALinear(in_features=4096, out_features=4096, r=16, lora_alpha=32.0)

# 3. Distributed Randomized SVD / Stateful Power-Iteration PCA
U, S, Vt = distributed_pca_svd(embeddings, k=32, method="randomized")
PYPI
matrix-compute-mcp
Install Command
pip install matrix-compute-mcp