wafergraph for AI agents (MCP server)

wafergraph is available as a remote MCP server, so an AI assistant can query the live dataset instead of relying on what it memorized. 30 read-only tools cover company search, structured screening and profiles, the 12-segment taxonomy and its subsegments, supplier and customer graph walks up to two tiers, documented supply paths between any two companies, disruption simulation, single-source dependency screening, the 74-deal M&A corpus and where consolidation is happening, side-by-side comparison of companies and of countries, country concentration per segment, chokepoint ranking weighted by market position, shared-upstream exposure for a list of tickers, and a dataset-statistics tool that reports coverage and freshness so an agent can see what the data cannot answer. No API key, no signup, no cost. Connect in Claude Code with: claude mcp add --transport http wafergraph https://mcp.wafergraph.com/mcp — or point any MCP client at that endpoint. It is open source (MIT) and listed in the official MCP registry as com.wafergraph/wafergraph-mcp. Responses carry their sources and disclose where coverage is partial rather than rounding the gaps away.