This is a notary layer for AI agents that need tamper-evident audit trails. It exposes five tools: mint a cryptographically timestamped record of any agent action, verify a single record by ID, list records with filters for agent, action type, and time range, fetch a full record including its original payload, and verify hash-chained sequences to confirm continuity. You'd reach for this when you have multiple agents making decisions or transactions and need proof that records weren't altered after the fact. Records live at evermint.app and can be verified publicly without an API key. The chain verification is especially useful for multi-step workflows where each action references the previous one.
claude mcp add --transport stdio evermint-app-evermint-mcp uvx evermint-mcp