neuralbrok 3.0.1

Agent orchestration platform — hardware-aware routing, MCP server, swarm intelligence, OpenAI-compatible

🧠 NeuralBroker 3.0: Unified Neural Gateway & Agent Orchestration

VRAM-Aware Routing · Knowledge Graph Memory · Swarm Intelligence · Zero-Trust Federation

---

⚡ The Ultimate Agent Orchestration Platform

NeuralBroker 3.0 transforms your machine into a Swarm-Native Orchestration Node. It goes far beyond standard model routing by merging the inference gateway, the swarm orchestrator, and structural memory into a single, self-improving daemon.

It automatically evaluates your hardware, coordinates multi-agent task pipelines, builds a persistent graph memory of its actions, and securely talks to other machines across the internet via Zero-Trust Federation.

pip install neuralbrok
neuralbrok start

That's it. Point Cursor, Claude Code, Cline, or any OpenAI/Anthropic-compatible app at http://localhost:8000, and the NeuralBroker Gateway takes over.

---

🌟 What's New in v3.0: The Unified Vision

🔌 1. Unified Inference Gateway

NeuralBroker now acts as a native Anthropic & OpenAI compatible proxy via a single port (:8000).

• Protocol Normalization: Transparently routes both /v1/messages (Anthropic) and /v1/chat/completions (OpenAI) calls to any provider (Ollama, NVIDIA NIM, DeepSeek, OpenRouter) using LiteLLM.
• Hardware-Aware Routing: Automatically falls back to cloud when VRAM is saturated or when task complexity requires a frontier model.
• Subscription Inheritance: It reads the OAuth session your Claude Code CLI holds in ~/.claude/.credentials.json and uses the claude binary as a free inference backend for hard tasks that spill over from your local GPU.

🕸️ 2. Neural Knowledge Graph & AgentDB

Every interaction creates a trace. NeuralBroker doesn't just store logs; it builds a persistent graph memory (NetworkX + SQLite) above its zero-dependency HNSW vector store.

• Entity Linking: Decisions, task outcomes, errors, and reasoning chains are stored as typed nodes.
• Auto-Learning: Agents query the graph before execution to inject historical context and avoid repeating past mistakes.
• Live Visualization: A real-time D3.js dashboard at http://localhost:8000/dashboard/graph visualizes the growth of your system's knowledge.

🛠️ 3. One-Command IDE Auto-Wiring

Zero configuration for your favorite IDEs. When you run neuralbrok start, NeuralBroker automatically detects your IDEs (Cursor, VS Code, Claude Code) and auto-wires them to use the local gateway by injecting the proper environment variables and settings.

🐝 4. Agentic Swarms & Task Coordination

NeuralBroker doesn't just route prompts; it routes objectives. The built-in SwarmCoordinator decomposes complex user requests into Plan → Execute → Review pipelines, automatically picking the best specialized local agent (Coder, Planner, Reviewer, Analyst) for each subtask.

🧠 5. NeuralFit Hardware Intelligence

Powered by a native Python implementation of our advanced hardware scoring algorithm, NeuralBroker scores models across Quality, Speed, Fit, and Context based on your exact hardware specifications (NVIDIA, Apple Silicon, AMD, or CPU). Run neuralbrok setup to see your live VRAM projections and auto-select optimal models.

🛡️ 6. Security & AIDefence

Every prompt is scrubbed:

• PII Redaction: AWS keys, emails, and SSNs are automatically masked before reaching cloud LLMs.
• Injection Shield: Federated incoming requests are heuristically scanned for adversarial jailbreaks.

🌐 7. Zero-Trust Federation

Swarms can securely communicate across networks. Using our local mTLS/Ed25519-style crypto module, your agents can encrypt and sign payloads and send them to other NeuralBroker nodes. Untrusted nodes are automatically downgraded.

🤖 8. Background Workers (Autopilot)

When your machine is idle, NeuralBroker wakes up. The internal WorkerDaemon launches background tasks to optimize codebases and run simulated security audits while you sleep.

🔌 9. MCP Server & Dynamic Plugins

NeuralBroker functions as a native Model Context Protocol (MCP) server via stdio. Connect Claude Code directly to it. You can dynamically load .yaml agent definitions into your ~/.neuralbrok/plugins/ directory and share them with the community.

---

💻 Quickstart

1. Install & Configure

pip install neuralbrok

# Detect hardware and auto-configure optimal models
neuralbrok setup

2. Start the Gateway

# Starts the unified gateway on port 8000 and auto-wires your IDEs
neuralbrok start

3. Explore the Knowledge Graph

# Open the live memory visualizer in your browser
neuralbrok graph open

# Query the structural memory
neuralbrok memory search "database schema"

4. View Telemetry Dashboard

Open http://localhost:8000/dashboard to view the live Pink Matrix telemetry UI, complete with routing waterfalls and VRAM gauges.

---

🏗️ Architecture

graph TD
    User([IDE / Agent SDK / Claude Code]) --> Gateway[NeuralBroker Gateway :8000]
    
    Gateway --> Policy[VRAM Policy Engine]
    Policy --> Local[Local Ollama / CUDA]
    Policy --> Cloud[Cloud / NVIDIA NIM / OpenRouter / Claude Subprocess]
    
    Gateway --> Swarm[Swarm Orchestrator]
    
    Swarm --> Memory[(Neural Knowledge Graph)]
    Memory --> Reasoning[Graph-Augmented Context]
    Reasoning -.-> Swarm

---

🛠️ Development

git clone https://github.com/khan-sha/neuralbroker.git
cd neuralbroker
python3 -m venv .venv && source .venv/bin/activate
pip install -e .
pytest tests/

📜 License

MIT © 2026 NeuralBroker Team. Built with 💖 by the Google Deepmind team (Advanced Agentic Coding).