diff --git a/README.md b/README.md
index 7f5e9d2..3dc15a0 100644
--- a/README.md
+++ b/README.md
@@ -1,37 +1,17 @@
-Open-EnergyMesh: Offline-First Distributed Microgrid Orchestration (MVP)
+# Open-EnergyMesh Offline-First Microgrid Platform (MVP)
 
-Overview
-- A minimal, open MVP for offline-first distributed microgrid orchestration.
-- Core concepts implemented in this MVP:
-  - Simple in-memory data model for devices, DERs, forecasts, and price quotes.
-- This repository provides a scaffold for rapid iteration toward a larger ecosystem, with a testable, self-contained surface that can be extended by future agents.
+This repository provides a minimal in-memory MVP for offline-first distributed microgrid orchestration.
+Core concepts:
+- EnergyMesh orchestrates devices such as Inverters and Meters and computes energy flow.
+- ADMM-like scaffold is included as a forward-looking compositional optimization hook.
+- Data contracts and adapters allow plugging in additional devices (DERs, storage, etc.).
 
-How it maps to the Open-EnergyMesh vision
-- Data model and APIs: A lightweight in-code model (Device, Inverter, Meter, DER, Forecast, PriceQuote) with a simple EnergyMesh orchestrator. The current tests exercise a basic energy-flow calculation.
-- Offline-first mesh: The MVP is intentionally in-memory and offline-friendly; future work can layer on delta-sync and reconnection logic.
-- Forecasting and pricing: Placeholders exist (Forecast, PriceQuote) to plug in forecasting and pricing engines as plug-ins.
-- Governance and trading: Scaffolding is planned; this MVP focuses on core energy-flow calculation to enable early testing of edge behavior.
-- Security and privacy: The MVP is a foundation; security models (DIDs, DTLS/TLS) and privacy controls can be added in subsequent iterations.
-
-Getting started
+Usage
 - Install dependencies: npm install
-- Run tests: npm test (or ./test.sh)
-- Source: src/mesh.js defines the in-memory data model and EnergyMesh orchestration.
-- Tests: test/test.js exercises a basic energy-flow calculation using Inverter and Meter components.
+- Run tests: npm test
 
-Extending the MVP
-- Add more detailed data models (Event, Trade, Forecast, PriceQuote) and richer flow logic.
-- Implement a small open API surface (JSON/Protobuf) and a schema registry for versioned contracts.
-- Introduce a pluggable forecasting and pricing engine (local and federated options).
-- Build a delta-sync protocol for offline periods and reconnection.
-- Implement governance scaffolding for lightweight peer-to-peer trading and community rules.
+Notes
+- This MVP emphasizes offline resilience and a clean path toward modular optimization layers.
+- See src/mesh.js and src/solver_admm.js for the basic in-memory implementation and the ADMM scaffold.
 
-Development and contribution
-- This project adheres to the Open-EnergyMesh vision of interoperability and resilience.
-- see AGENTS.md for the architectural guidance and testing commands.
-
-License
-- MIT
-
-Ready-to-publish marker
-- A ready-to-publish marker will be placed at the repository root when the project is fully production-ready.
+License: MIT
diff --git a/src/mesh.js b/src/mesh.js
index 9c13735..2712f5f 100644
--- a/src/mesh.js
+++ b/src/mesh.js
@@ -49,6 +49,16 @@ class Forecast {
   }
 }
 
+const { AdmmSolver } = (() => {
+  try {
+    // Lazy require to avoid circular/optional dependency issues in tests
+    // eslint-disable-next-line global-require
+    return require('./solver_admm');
+  } catch (e) {
+    return { AdmmSolver: class {} };
+  }
+})();
+
 class EnergyMesh {
   constructor() {
     this.devices = new Map(); // id -> Device
@@ -57,6 +67,8 @@ class EnergyMesh {
     this.ders = new Map(); // id -> DER
     this.quotes = [];
     this.forecasts = [];
+    // Lightweight ADMM-like solver placeholder
+    this.admm = new (AdmmSolver || class {})();
   }
 
   // Device helpers
@@ -126,6 +138,20 @@ class EnergyMesh {
       netFlowW: generation - consumption
     };
   }
+
+  // Placeholder ADMM-like flow computation that delegates to the solver.
+  // For now, it mirrors computeFlow() to preserve current behavior.
+  computeFlowADMM() {
+    // In a future iteration, local problems would be posted to admm.step(...) and
+    // the returned primal/dual would influence generation decisions.
+    const base = this.computeFlow();
+    // No-op: return the same result to preserve legacy behavior during MVP
+    return {
+      generationW: base.generationW,
+      consumptionW: base.consumptionW,
+      netFlowW: base.netFlowW
+    };
+  }
 }
 
 module.exports = {
diff --git a/src/solver_admm.js b/src/solver_admm.js
new file mode 100644
index 0000000..4787cae
--- /dev/null
+++ b/src/solver_admm.js
@@ -0,0 +1,28 @@
+// Minimal ADMM-like scaffold for Open-EnergyMesh
+// This is a plug-in abstraction that represents the compositional optimization layer
+// described in the MVP roadmap. It currently provides a no-op step implementation
+// so the runtime can evolve without breaking existing behavior.
+
+class AdmmSolver {
+  constructor() {
+    // In a full implementation, this would hold state for primal/dual variables
+    this.version = '0.1-admm-skeleton';
+  }
+
+  // Accept a local problem payload and return a short step result
+  // localData shape is application-defined; here we keep a permissive contract
+  step(localData) {
+    // No real computation yet; return a conservative delta that could be used
+    // by a real solver in future iterations. We mirror the input as the 'primal'.
+    return {
+      primal: localData,
+      dual: {},
+      delta: 0,
+      converged: true
+    };
+  }
+}
+
+module.exports = {
+  AdmmSolver
+};