Stellar Raven is an MCP server that connects AI agents to live Stellar ecosystem data—documentation, news, repositories, and community intelligence. Tyler Vanderhoeven from SDF explains how the tool prevents agents from canonicalizing outdated code patterns through a single integrated entry point.
Stellar Raven, a new MCP (Model Context Protocol) server from SDF's Frontier Lab, connects AI agents to comprehensive, scored Stellar ecosystem data through a single entry point. Presented by Tyler Vanderhoeven at Meridian 2026, Raven aggregates official documentation, community intelligence from Lumenloop, builder data from Stellarlight, and ecosystem tools to ground agents in current best practices. The core problem Raven solves: when AI agents rely on outdated information—like older Soroban contract patterns—they reproduce those patterns in new code, making the old way canonical again. Raven's solution uses the agent's own reasoning to decide which services to query and validate answers, avoiding expensive model-in-service approaches. Early adoption is strong: 600 users have already asked over 1.5 million questions. The tool is positioned for builders, researchers, autonomous monitoring, and discovery, helping agents avoid ecosystem confusion while exploring what already exists before building something new.
