Build

MCP

Verified data as a tool your agent can call.

Metera runs a Model Context Protocol server at https://api.metera.xyz/mcp (protocol 2024-11-05). Once connected, your agent can ask for what it needs in plain language or with a precise request for a specific type, the same way it would ask a person. What comes back is exactly the delivery object the HTTP API returns: the datum, its verification level, the source, and a trace id, as JSON text. There is nothing extra to parse or trust differently because the connection happened over MCP instead of HTTP.

The catalog itself is discoverable from inside the connection, so an agent can learn what is servable today, and what each type requires, without leaving the conversation.

One tool goes further than fetching a datum: evaluate_gate composes a token risk check and a real simulated swap quote into an allow / allow_with_caveat / refuse decision — against a policy your agent supplies, never a threshold Metera picked. See the Gates page.

Claude Code (tested here)

claude mcp add --transport http metera https://api.metera.xyz/mcp --scope user

Cursor and Codex: both support HTTP MCP servers. Point them at https://api.metera.xyz/mcp with a Bearer mk_live_… header, per each tool official MCP configuration. Those two setups follow the vendor documentation and were not tested here. Only the Claude Code path and the raw MCP endpoint were run in verification.

PreviousAPI reference

NextGates