Trende: Multi-Source AI Research with Cryptographic Proof

Trende researches any topic across social media, news, forums, and market data using multiple AI models, then produces cryptographically verified research reports.

What it does:

1. Research: Multi-platform analysis across 17 data sources via LangGraph agents
2. Verify: Every report is TEE-attested with cryptographic signatures (Eigen)
3. Distribute: Publish to Paragraph, settle on-chain via Chainlink, or serve via A2A API
4. Automate: ACP integration lets other AI agents hire Trende for research

Live deployments: Arbitrum Sepolia + Base Sepolia ✅

📖 Documentation

Consolidated documentation (max 4 docs, 300 lines each):

• Architecture: System overview, LangGraph workflow, TEE attestation, Chainlink oracle
• Integration: ACP, Paragraph, Chainlink integration guides
• API Reference: Complete API documentation for developers
• Developer Guide: Quick start, deployment, troubleshooting

🚀 Quick Start

Backend (Python 3.10+)

python -m venv venv
source venv/bin/activate
pip install -e .
# Configure .env based on .env.example
uvicorn backend.api.main:app --reload

Frontend (Node.js 18+)

cd frontend
npm install
npm run dev

Testing

# Run a live research task
python3 scripts/test_agent.py "Your Research Topic"

# Run full flow (start -> poll -> Forge links)
./scripts/finals_flow.sh "Your Research Topic"

🤖 Agent Ecosystem (A2A)

Trende is built for the agent-to-agent economy:

• llms.txt: Master Discovery File for LLM-based agents.
• Verifiable Output: Every analysis produces an Attestation Payload signed by a TEE.
• Settlement: Native X402 (EIP-3009) support for automated intelligence purchases.
• Agentic UX: Live "Deploy Agent" dispatch, visible on-chain oracle state, autonomous sentinel-triggered settlement, and copyable A2A invocation payloads.

🛡️ Core Principles

• Verifiable First: Cryptographic proof for every report, leveraging Eigen TEE for verifiable compute.
• Multi-Model: Eliminates single-source bias using Venice, GPT-4o, Llama, and Gemini.
• Chain Agnostic: Built to serve intelligence to any ecosystem, including Monad, Base, BNB, and Solana.
• Privacy Centric: Primary inference routed via Venice AI.

🔗 Chainlink Integration

Trende uses Chainlink Functions and Chainlink Runtime Environment (CRE) for verifiable social intelligence.

• Contracts: /contracts (✅ live oracle receiver path on Arbitrum Sepolia + Base Sepolia)
• Functions: JS sources in backend/chainlink/functions
• CRE Workflow: Decentralized consensus in backend/chainlink/cre/workflow/

Verifiable Features

1. Data Sourcing: GDELT & CoinGecko via Chainlink Functions
2. Oracle: TrendeOracle implements Chainlink IReceiver for CRE-delivered settlement
3. Autonomous Settlement: Sentinel loop auto-triggers Functions resolution when markets mature
4. CRE Status: receiver path is live and simulation is verified on Arbitrum Sepolia; workflow deployment is pending Chainlink deployment access approval

Current Arbitrum Sepolia Contracts

• TrendeOracle: 0xEEDeD7daC9D6b17f5D3915542A549B1AefCeed56
• TrendeFunctionsConsumer: 0xA4C4FC79909165fFeAFEdEb47A93Db058383DB84
• Verified CRE simulation trigger tx: 0xcad4b3455e9d53281d6393318272eb01b98311740abbcae393d738829b93a3e0

✅ Production

• CI gates: /.github/workflows/ci.yml
• Smoke tests: scripts/smoke_matrix.sh

---

Monad Testnet: Chain ID 10143 | RPC https://testnet-rpc.monad.xyz | Explorer