Clay MCP

Clay's MCP server exposes go to market data and enrichment logic to AI clients, so a rep or an agent can build a prospect list without opening the product. People search finds contacts by attributes such as title, role, department, company and location. Company search queries Clay's company data by industry, technology used, funding and size. Enrichment runs the waterfall model Clay is built on, checking a request against its network of data vendors in sequence until a validated email or phone number comes back, rather than trusting a single provider. Saved enrichment and scoring logic a team has already built can be invoked, so an agent applies the same qualification rules as the rest of the organization. Table access covers listing workspace tables and reading their data, which lets an agent work against an existing audience instead of starting from scratch. On the write side, enriched contacts can be pushed to a CRM with duplicate checking, and contacts can be sent into an outreach sequencer, with operations teams defining the write rules and compliance logic those actions must follow. There are two shapes of access. A hosted remote endpoint works with chat style clients that support remote MCP connections. A local path runs the Clay CLI in MCP mode, authenticating from credentials established by a CLI login instead of an API key pasted into the host config, which suits coding agents such as Claude Code, Codex and Cursor that can run a local process and execute commands.