Ambr

This connects Claude to Ambr's permission and contract infrastructure for AI agents. You get tools to create authorization boundaries (spend caps, time windows, counterparty whitelists), mint Ricardian contracts as cNFTs on Base L2, and verify actions cryptographically. The dual format approach means every contract is both human-readable (Markdown/PDF for lawyers) and machine-parsable (strict JSON for agent execution), bound together by SHA-256 hash. Reach for this when you're building agents that spend money, delegate tasks to other agents, or need an auditable record of decisions that holds up outside the prompt window. Integrates with x402 for payments and A2A protocol for agent discovery.