This connects agent tool calls to a verification layer that checks operational evidence before actions execute. When your agent calls a tool, GeoClear sits in front as MCP-compatible middleware and returns one of four outcomes: valid evidence proceeds, missing evidence holds with structured guidance, policy violations block, tampered evidence rejects. You keep raw mission data, PII, and logs inside your boundary while the system verifies signed evidence packets locally without live API calls. Three deployment modes: customer-computed policy results that GeoClear signs, minimized attribute evaluation, or full enclave deployment for federal and high-assurance environments. Reach for this when you need auditable proof of what your agents attempted, not just what they returned.
claude mcp add --transport http io.github.shaileshjgd-geoclear https://mcp.geoclear.io/mcp