Paste a token address, get a readable verdict. On-chain data through risk heuristics, then a fine-tuned model.
Switch between how the work ran before, and how it runs now.
In the order it was built, and why each part exists.
A chat dashboard rather than a report generator: paste an address, get a verdict, ask a follow-up. Two response modes, Standard and Psycho, for people who want the analysis blunt.
A submitted token triggers an n8n workflow that pulls live on-chain data, rather than an application calling an API and hoping.
Holder concentration, liquidity, contract issues and suspicious transfer patterns, evaluated consistently for every token so two checks are comparable.
The evaluation is fed to a fine-tuned model that turns it into a human answer inside the same conversation, instead of leaving the user to interpret raw numbers.
On-chain data is public and almost useless to a non-specialist. The information that tells you whether a token is dangerous is spread across explorers, and it takes experience to weigh it.
The gap was never access to data. It was interpretation.
A conversational dashboard with an automation backend. The interface is deliberately plain: an address goes in, an analysis comes back, and the conversation continues.
Behind it, an n8n workflow gathers live chain data, runs it against risk heuristics, and hands the result to a fine-tuned model for the write-up. Ethereum, BSC, Polygon, Solana and PulseChain are supported.
Putting the analysis in a workflow rather than in application code means a heuristic can be changed, retried or rerun without redeploying the product.
It also means each stage is inspectable. When a verdict looks wrong, you can see which step produced it.
5
Chains supported
2
Response modes
1
Input: the token address