Skip to main content
Blockgence is an on-chain intelligence platform for investigating funds on EVM networks. It traces the path of funds, attributes addresses to known entities, runs explainable risk screening and analyzes contracts and incidents. Every result says where it comes from and what could not be seen. There are three ways to use it, and all three share your account, your cases and your credit quota:

Web app

The investigation canvas, cases, Monitor, Guard and Audit, with your user.

REST API

The same queries from your code, under /v1, with an API key.

MCP

Connect your AI assistant (Claude, Cursor, Codex…) and query from the chat.

What you can query

  • Fund tracing: the transfer graph from an address or from the transaction of a theft, including bridge crossings and swaps.
  • Attribution and screening: entity labels, OFAC sanctions with their history and an explained risk score.
  • Contracts and tokens: who controls a contract, transaction simulation and a token profile.
  • Incidents: the decoded exploit transaction, approvals and signature drains, the compromise vector, freeze status with stablecoin issuers and the evidence package.
  • Cases and evidence: save an investigation, share it with your organization and export a sealed, verifiable package.
Ethereum, BNB Chain, Polygon, Arbitrum, Base, Optimism and, on the functions that say so, Tron are queried. The exact list, with what can be queried on each one, is returned by GET /v1/chains (free). It also lists networks that are only recognized and not read (queryable: false), such as Monero or Bitcoin.

First steps

1

Request an account

Access is by invitation. Request one from the website and, once it is approved, you get a link to create your user. See Accounts and organizations.
2

Create a key

In the app, your name’s menu → MCP… → Crear clave (create key). The same key works for the API and for the MCP. See API keys.
3

Make your first request

It returns your plan, your limits and what you have used this month. It costs no credits.

How responses work

If a data source fails, the response says so. If what you asked for could not be read, you get a 503 and are not charged; if an intermediate step failed, the response still arrives, with degraded_sources. See Errors.
Reads have caps. Whatever was left unread travels in the response, with the missing range and how to ask for more. See Pagination and cutoffs.
An on-chain fact and a heuristic are not worth the same. Every claim states its basis and its confidence; leads are labeled as leads. The full methodology is published and can be queried through the API.
An error costs no credits. See Credits and limits.