Overview Advanced APIChains list Verifiable RPC Agent RPC Free Credits

Blockchain data your agent can read,
not just fetch.

Teams building on Ankr:

Point a coding agent at one MCP endpoint, or add a single header to your own JSON-RPC calls. Either way the answer arrives ABI-decoded, hex turned into decimals and the bulk dropped. A transaction receipt we measured came back at 1,177 bytes instead of 3,182.

Teams building on Ankr:

Three ways to give an agent chain data.

Agent RPC MCP

A hosted Model Context Protocol server with seventeen read-only tools. Point Claude Code, Cursor or VS Code at one URL, authenticate with the API key you already have, and the agent reads transactions, logs, balances and contracts. Nothing in it can sign or broadcast.

TORPC, one request header

Ask for a decoded response on ordinary JSON-RPC and event logs come back as named arguments, hex as decimals, and logsBloom gone. It is an open specification under CC0, so any provider can implement it and any HTTP client can use it.

Verifiable RPC

Every response is signed inside an Intel TDX enclave and the proof travels with it. Your client verifies locally and fails closed, so an agent never acts on data it could not check.

The first two are Agent RPC.

The header opt-in and the hosted MCP server are one product: Agent RPC. Add the Accept-Token-Tier header to JSON-RPC calls you already send, or point a client at the hosted MCP server and its seventeen read-only tools using the API key you already have. Same endpoints, same rate limits, nothing to migrate.

See how Agent RPC works