Morpheus Lumerin Node

May 23, 2026 · View on GitHub

Simple-Overview

The purpose of this software is to enable interaction with distributed, decentralized LLMs on the Morpheus network through a desktop chat experience.

v7.0.0 — Full TEE capability. The v7 release completes a two-hop Trusted Execution Environment (TEE) trust chain for any model registered on-chain with the tee tag:

  • Phase 1consumer → P-Node. A consumer proxy-router (v6.0.0+) cryptographically verifies the provider's P-Node runs the exact official hardened -tee image inside a genuine Intel TDX SecretVM, with TLS pinning, at session open and on every prompt.
  • Phase 2 (new in v7)P-Node → backend LLM. The v7+ P-Node itself verifies the backend LLM it forwards inference to (CPU TDX quote, TLS pinning, workload RTMR3 replay of the backend's docker-compose.yaml, CPU-GPU nonce binding, and NVIDIA NRAS GPU attestation) at startup and on every prompt.

Because Phase 2 runs inside the attested P-Node, any v6+ consumer is forward-compatible with a v7+ provider and gains the Phase 2 guarantees automatically — no client-side upgrade required. See the TEE reference, the SecretVM quickstart, and the developer reference at proxy-router/docs/tee-backend-verification.md.

Documentation

The canonical documentation lives at nodedocs.mor.org. Source files are in /docs and built with Mintlify. The site replaces the previous 00-overview.md / 02-*.md / 04-*.md / 99-troubleshooting.md set of files; old paths still resolve via redirects in docs/docs.json.

The site is structured around role-based journeys (consumer / prosumer / provider tiers), with anti-hallucination AI knowledge pages and curated mirrors of the broader ecosystem (mor.org, tech.mor.org, active.mor.org, MyProvider, Everclaw, NodeNeo, app.mor.org).

What's in this repo

  • Local Llama.cpp and tinyllama model to run locally for demonstration purposes only.
  • Lumerin proxy-router — background process that monitors blockchain contract events, manages secure sessions between consumers and providers, and routes prompts and responses between them.
  • Lumerin MorpheusUI — the Electron front end UI to interact with LLMs and the Morpheus network as a consumer.
  • Lumerin cli — CLI client to interact with LLMs and the Morpheus network as a consumer.
  • Kubo ipfs — IPFS client to store and retrieve model/agent files.

End-to-end picture

  1. PreRequisites: BASE Layer 2 Blockchain, MOR and ETH on BASE for staking and bidding.
  2. Existing, Hosted AI model available for inference via the Morpheus network.
  3. The proxy-router talks to and listens to the blockchain; it routes prompts and inference between providers' models and consumers.
  4. Providers register their models via bids on the blockchain.
  5. The consumer node is the "client" that purchases bids, sends prompts via the proxy-router, and receives inference back from the provider's models.
  6. Consumers stake MOR to open a session for the duration they intend to use.
  7. Once the session is open, prompt and inference (ChatGPT-like) can start.

Tokens and contract information

ItemBASE MainnetBASE Sepolia (testnet)
Chain ID845384532
Branchmain (MAIN-* releases)test (*-test releases)
MOR Token0x7431aDa8a591C955a994a21710752EF9b882b8e30x5C80Ddd187054E1E4aBBfFCD750498e81d34FfA3
Diamond Marketplace0x6aBE1d282f72B474E54527D93b979A4f64d3030a0x6e4d0B775E3C3b02683A6F277Ac80240C4aFF930
Block Explorerhttps://base.blockscout.com/https://base-sepolia.blockscout.com/

You will need both MOR (for stake / fees / session payment) and ETH on BASE (for gas) in your wallet.

Quickstart

RoleStart here
Consumer (chat)Consumer quickstart
Provider (host your own model)Provider quickstart
TEE provider (SecretVM)SecretVM quickstart
Resale providerResale overview
Prosumer / agentProsumer overview
Developer (API)API overview

For AI agents reading this repo

Start with AGENTS.md — hard rules, quick lookup tables, and ingestion instructions.

To load the full documentation corpus in one fetch:

ResourceURLUse
Full corpus (preferred)llms-full.txtComplete markdown export — fetch this, do not scrape HTML pages
Page indexllms.txtTitles, descriptions, and slugs for every page

Individual nodedocs page URLs are for human browsing and citation only. See AGENTS.md for priority reading slugs and anti-hallucination rules.