Orynt AI is a sophisticated Proof-of-Concept (POC) blockchain agent designed to transform natural language queries into verified on-chain insights. It leverages the Gemini 3 Pro reasoning engine and Etherscan v2 authenticated infrastructure to provide a "Think-Before-Act" intelligence workflow.
Orynt AI follows a strict 6-step lifecycle to ensure data accuracy and user trust:
- Intent Parsing: The user's query is analyzed to extract the Subject (e.g., Ethereum), Action (e.g., Cost Estimation), and Urgency. This prevents "hallucinated actions" by anchoring the request to a known schema.
- Data Planning: Instead of blindly calling APIs, the agent formulates a Data Strategy. It selects specific tools (e.g.,
getMempoolStats+getEthMarketData) required to satisfy the specific intent. - Tool Execution: The agent executes authenticated calls to the Etherscan v2 API. This layer includes robust hex-to-decimal conversion and error handling for sub-gwei environments.
- Aggregation: Raw JSON-RPC responses and Oracle data are normalized. We calculate the
pressureIndex(Node Price / Oracle Price) to measure real-time mempool congestion. - Insight Generation: Gemini synthesizes the collected data into a human-readable report. This includes specific USD costs (formatted as
< $0.01for low-gas environments) and uncertainty scores. - Grounding & Sources: Every insight is linked back to its origin (Etherscan v2), providing a transparent "Chain of Custody" for the intelligence.
- Gemini 3 Pro Reasoning: Advanced semantic understanding of complex blockchain queries.
- Etherscan v2 Integration: Uses the latest v2 API for higher performance and broader chain support.
- Execution Control: Includes an
AbortControllermechanism allowing users to stop the reasoning process mid-flight. - Sub-Gwei Precision: Specifically tuned to handle extremely low gas environments with high decimal precision.
- Live Console: A terminal-style view that provides transparency into the agent's internal "thought process" and API interactions.
- Read-Only: The agent has no "Write" permissions. It cannot sign transactions or access private keys.
- Authenticated Traffic: All data is fetched via authenticated Etherscan endpoints to ensure reliability and higher rate limits.
- Context Isolation: Each session is grounded only in the data retrieved during that specific execution cycle.
- Node.js (v20 or higher)
- npm or yarn package manager
- Gemini API Key - Get it from Google AI Studio
- Etherscan API Key - Get it from Etherscan
-
Clone or download the repository:
cd blockchain-intelligence-poc -
Install dependencies:
npm install # or yarn install
Orynt AI implements a secure BYOK (Bring Your Own Keys) model where API keys are never exposed or stored on servers:
- Launch the application - On first visit, a settings modal will prompt you to enter your API keys
- Enter your keys:
- Gemini API Key: Get from Google AI Studio
- Etherscan API Key: Get from Etherscan
- Keys are stored securely:
- Stored in browser
sessionStorage(cleared when tab is closed) - Never sent to any backend server
- Never logged or persisted to disk
- Stored in browser
- Update keys anytime - Click the "Keys" button in the header to reconfigure
Security Benefits:
- 🔒 Frontend-only key management
- 🚫 No backend exposure
- 🔄 Session-scoped (auto-clear on tab close)
- 👤 Complete user control
Development Mode:
npm run dev
# or
yarn devThe application will start on http://localhost:3000. A settings modal will appear for API key configuration.
Build for Production:
npm run build
# or
yarn buildThis creates an optimized build in the dist/ directory.
Preview Production Build:
npm run preview
# or
yarn previewThis serves the production build locally for testing.
- Start the application - The Settings modal will appear
- Configure your API keys - Enter your Gemini and Etherscan keys (BYOK model)
- Query the blockchain - Enter a natural language query about Ethereum data
- Review the results - Watch the live console for the agent's reasoning process
- Explore insights - See synthesized data with sources and confidence metrics
This project is optimized for Netlify with a preconfigured netlify.toml file.
-
Push to GitHub (or your Git provider):
git add . git commit -m "Initial commit" git push origin main
-
Connect to Netlify:
- Go to Netlify
- Click "New site from Git"
- Connect your repository
- Build command:
npm run build - Publish directory:
dist - Deploy!
-
Configure Environment Variables (if needed):
- In Netlify dashboard, go to Site settings → Build & deploy → Environment
- Note: API keys are configured in the browser UI (BYOK), no backend env vars needed
The netlify.toml file automatically handles:
- ✅ Correct MIME types for JavaScript modules
- ✅ Single-page app routing (SPA)
- ✅ Security headers
- ✅ Proper asset serving
Powered by Google Gemini 3 Pro & Etherscan v2 Infrastructure.
