Inspiration
Blockchain security investigations can require analysts to examine smart contracts, transactions, wallets, liquidity, and historical activity across multiple sources. We wanted to explore whether a team of AI agents could work together to investigate these signals continuously instead of relying entirely on manual analysis.
This led to ClawIntel — a system designed to stream blockchain intelligence as it unfolds while also allowing users to submit a specific token or contract address for an on-demand investigation.
What it does
ClawIntel is an autonomous multi-agent blockchain security intelligence platform.
It continuously streams blockchain intelligence as events unfold and uses specialized agents to investigate different aspects of tokens, smart contracts, transactions, and deployer activity.
Users can also paste a token or smart-contract address into ClawIntel to trigger a targeted investigation. The agents analyze the target, correlate their findings, and produce security intelligence that can be observed through the platform in real time.
How we built it
We built ClawIntel around a multi-agent architecture where specialized agents handle different parts of a blockchain investigation and a central intelligence layer combines their findings.
The system connects to blockchain data sources, APIs, RPCs, market data, and other intelligence sources. We use multiple LLM providers — OpenRouter, Google Gemini, and Cerebras — to support agent reasoning, investigation, interpretation, and evidence synthesis.
The platform was built primarily with Python, with a web-based interface using JavaScript, HTML, and CSS, and MongoDB for persistent data.
Challenges we ran into
One of our biggest challenges was coordinating multiple agents while keeping their investigations consistent and useful in real time.
We also had to deal with the complexity of combining blockchain data from different networks and sources while making the information understandable to users as it was being generated.
Another challenge was balancing AI reasoning with deterministic analysis so that the system could reason over real technical evidence instead of relying entirely on an LLM response.
Accomplishments that we're proud of
We are proud of building a system where blockchain intelligence is not simply displayed after an analysis is complete — the investigation itself can be streamed as it unfolds.
We are also proud of combining continuous monitoring with on-demand forensic investigation. A user can observe live intelligence or provide a specific token or contract address and let the agents investigate it.
Building a multi-agent architecture that brings together contract analysis, behavioral investigation, deployer intelligence, and AI-powered evidence synthesis was another major accomplishment.
What we learned
We learned that effective AI agents need more than an LLM. They need reliable data, structured workflows, specialized responsibilities, and mechanisms for connecting evidence.
We also learned that blockchain security investigation is highly multi-dimensional. Contract behavior, transactions, wallets, liquidity, and historical activity can each reveal different pieces of the same security picture.
Most importantly, we learned how multiple AI models and specialized agents can work together to turn large amounts of constantly changing blockchain data into more useful security intelligence.
What's next for ClawIntel
We plan to expand ClawIntel into a broader autonomous blockchain security intelligence platform.
Future development includes deeper smart-contract vulnerability analysis, stronger cross-chain investigation, more advanced threat-pattern detection, continuous monitoring of specific contracts and wallets, richer threat intelligence, and integrations with professional security workflows.
Our long-term goal is to make ClawIntel a system that can continuously investigate blockchain activity, surface meaningful security evidence, and help human security researchers respond to emerging threats faster.
Built With
- blockchain-rpcs
- cerebras
- dex-screener
- gemini
- html/css
- javascript
- jupiter
- mongodb
- openrouter
- python
- websockets
Log in or sign up for Devpost to join the conversation.