09 · FOR BUILDERS
They shipped the hands. We shipped the eyes.
9. MCP: Intelligence for Trading Agents
9.1 Hands and Eyes
In 2026, Robinhood extended its agentic trading platform to crypto, giving AI agents the ability to execute trades through its Trading MCP. Execution is one half of an agent. The other half is knowing what to do, and that is the half FAUST serves. Robinhood’s MCP lets agents act; FAUST’s MCP lets agents see: live signals, wallet histories, token flows, and the complete indexed record of Robinhood Chain. The two are complementary and unaffiliated, with zero overlap.
9.2 Connecting an Agent
FAUST exposes its dataset over the Model Context Protocol, the open standard that lets AI systems query external tools. Any MCP-capable agent or platform, including Claude, ChatGPT, and open agent frameworks, can connect to the FAUST MCP server and query the dataset conversationally or programmatically. Connection details and the current server address are published at faustprotocol.xyz.
9.3 What Your Agent Can Query
Token intelligence: activity, windows, and flow for any token on the chain, resolved by address, with impostor disambiguation built in.
Chain state: live and historical volume, swaps, active wallets, and new-pool creation.
RWA data: the tokenized-stock index, pairs, and per-stock activity from Section 7.
Verification: the same windows FAUSTeyes posts, recomputable on demand.
9.4 Use Cases
A trading agent checks a token’s real indexed activity, and whether the ticker it was told about is the canonical asset or an impostor, before executing. A research agent pulls the RWA dataset to analyze the chain’s tokenized-stock economy. A monitoring agent watches windows around its own portfolio and reacts to verified flow rather than timeline sentiment.