Autonomous AI Governance Agent

An autonomous, stateful AI Agent for decentralized governance, engineered to intelligently evaluate DAO proposals and execute automated, on-chain voting.

The agent utilizes Retrieval-Augmented Generation (RAG) to parse incoming proposals against historical DAO constitutions, rules, and past decisions, calculates a dynamic risk assessment, and securely signs and broadcasts voting transactions to the Solana Devnet ledger.

---

🚀 Key Features

• Stateful AI Workflows: Orchestrated via LangGraph, utilizing a dedicated multi-node graph (Researcher -> Analyst) to handle complex cognitive reasoning loops.
• Context-Aware Analysis (RAG): Integrates LangChain Chroma and Ollama Embeddings to retrieve deep semantic context from past historical DAO data and local constitutional documents.
• Automated Risk Grading: Evaluates proposal logic to dynamically output risk scores, analytical logs, and explicit governance recommendations (YES, NO, ABSTAIN).
• On-Chain Vote Execution: Automatically drafts, signs, and pushes modern Solana VersionedTransaction payloads to the Solana Memo Program to permanently store vote records.
• Autonomous Key & Gas Management: Securely generates and caches local file-system keypairs and auto-requests airdrops from the Solana Devnet faucet if gas balances drop below minimum thresholds.

---

📂 Project Architecture

backend/
├── agent/
│   ├── database.py       # Local Chroma DB initialization & document seeding
│   ├── pipeline.py       # RAG logic for semantic document retrieval
│   └── workflows.py      # LangGraph state machine & evaluation workflow runner
├── web3/
│   ├── wallet.py         # Persistent local Solana Keypair storage & gas checks
│   └── governance.py     # Asynchronous versioned transaction signer & broadcaster
├── tests/                # Test suites tracking module behavior
├── config.py             # Pydantic-Settings environment validation setup
├── requirements.txt      # Requirement files for virtual env setup
└── .env                  # Configuration variables (git ignored)

---

🛠️ Tech Stack

• Frameworks: LangGraph, LangChain (Chroma, Ollama integrations)
• Vector Storage: Chroma DB (Local Persistent Storage)
• Inference Engine: Ollama (Local LLM and Embeddings Engine)
• Blockchain Integration: Solana Python SDK (solana, solders)
• Environment & Validation: Pydantic Settings, python-dotenv

---

⚙️ Installation & Setup

1. Prerequisites

Ensure you have the following installed locally:

• Python 3.10+
• Ollama (Make sure it is running locally with your chosen embedding/LLM models pulled down)

2. Clone and Install Dependencies

Navigate to your project root directory and install required Python libraries:

pip install langgraph langchain-chroma langchain-ollama solana solders pydantic-settings python-dotenv

3. Environment Configuration

Create a .env file inside the backend/ directory:

OLLAMA_BASE_URL=http://localhost:11434
EMBEDDING_MODEL=nomic-embed-text
CHROMA_COLLECTION_NAME=dao_governance_rules
CHROMA_DB_DIR=chroma_db
SOLANA_RPC_URL=https://api.devnet.solana.com
RPC_URL=https://api.devnet.solana.com
MEMO_PROGRAM_ID=Mem1111111111111111111111111111111111111111

---

🔌 Usage Guide

🧬 Step 1: Initialize & Seed the Vector Database

Populate your local Chroma vector store with constitutional rules, financial histories, and guidelines:

python -m agent.database

🤖 Step 2: Execute the AI Agent Workflow

To pass a new test proposal through the multi-node LangGraph analysis pipeline and run the automatic on-chain execution:

python -m agent.workflows

• How it works: The workflow automatically loads the internal project loop, triggers an internal RAG lookup against your Chroma DB vector store, evaluates risk metrics, generates a decision, handles keypair checking, and submits the finalized transaction to Solana Devnet.

---

🧪 Testing

To verify functionality across both core modules, run the unit test commands via pytest (if configured):

pytest tests/