Active 0G Mainnet Validator — this project is built and operated by an active validator on 0G mainnet (chain ID 16661). Validator address: 0xaED4832042D1204Faf7a97eDD93611A92B20461c

Live RPC connection verified. See build/live_run.log for the captured connection log.

Universal AI-powered ranking & indexing infrastructure for the 0G Network.

This repository contains the production-grade prototype that accompanies the DecentraRank Guild on 0G 2.0 grant proposal.

decentrarank/
├── INDEX_DECISION.md             # Why we index validators first
├── SCHEMA.md                     # Universal schema specification
├── README.md                     # This file
├── requirements.txt
├── decentrarank/
│   ├── schema.py                 # Canonical Entity dataclass
│   ├── signals.py                # Five universal ranking signals
│   ├── ingestion/
│   │   ├── rpc_client.py         # 0G JSON-RPC client (retries, backoff)
│   │   ├── validator_adapter.py  # 0G validator → universal Entity
│   │   └── pipeline.py           # End-to-end ingestion pipeline
│   └── mabs/
│       ├── agents.py             # 3-agent architecture
│       ├── simulation.py         # Discrete-step simulation engine
│       └── calibration.py        # Weight-calibration sweep
├── scripts/
│   ├── ingest_validators.py      # Run the ingestion pipeline
│   └── run_simulation.py         # Run MABS + calibration end-to-end
└── tests/
    ├── test_schema_and_signals.py
    └── test_ingestion.py

# 1. Install dependencies
pip install -r requirements.txt

# 2. Run the ingestion pipeline (replay mode — no network access required)
python scripts/ingest_validators.py --replay

# 3. Run the full MABS simulation + calibration
python scripts/run_simulation.py --replay --steps 100

# 4. Run the test suite
python -m unittest discover tests -v

To run against a live 0G mainnet endpoint, configure both the JSON-RPC and the Tendermint REST gateway endpoints:

export ZEROG_RPC_ENDPOINT="https://your-0g-rpc"
export ZEROG_REST_ENDPOINT="https://your-0g-rest-gateway"

python scripts/ingest_validators.py
python scripts/run_simulation.py --steps 200

If the RPC endpoint is unreachable or the REST gateway is not configured, the pipeline falls back gracefully to replay mode and clearly logs that it has done so.

Three layers, each leveraging 0G's modular infrastructure:

1. Universal Ingestion & Adapter Framework — domain-specific adapters translate any structured dataset into a single canonical schema. The validator adapter is the reference implementation. Adapters for DeFi, DeSci, and AI-agent domains follow the same pattern.

2. MABS Ranking Engine — three agent types (ProducerAgent, ConsumerAgent, ValidatorOfValidatorsAgent) interact under configurable rules, and ranking weights are derived from the emergent stability of those interactions rather than hand-coded.

3. Verifiable Persistence on 0G — entity records are content-addressable and designed to be persisted on 0G Storage with proofs in the 0G DA layer. This prototype produces JSON output ready for storage submission; the production daemon (Milestone 2) wires it to the live 0G stack.

This codebase is a working prototype at v0.1. The grant funds:

• Milestone 1 — schema lock at v1.0; production-grade adapters for two more verticals; full integration test suite.
• Milestone 2 — MABS engine deployed to production with the calibration loop running continuously; live integration with 0G Storage and DA.
• Milestone 3 — public REST + GraphQL query API; verifiable execution via 0G Compute; security audit.
• Milestone 4 — open-source release; deployment guide; research paper draft.

All code: Apache License 2.0 All documentation and research outputs: CC-BY-4.0

Hanumant Joshi — hanumantjoshi44@gmail.com

Validator address (active on 0G mainnet): 0xaED4832042D1204Faf7a97eDD93611A92B20461c