# Obsigna > Tooling for the Agent Receipts protocol Obsigna is the open-source toolset for producing and verifying Agent Receipts — cryptographically signed, hash-chained records of AI agent actions. It provides SDKs, an MCP proxy, a PostToolUse hook, a collector daemon, and a dashboard. ## Key links - Getting Started: https://obsigna.dev/getting-started/quick-start/ - MCP Proxy: https://obsigna.dev/mcp-proxy/overview/ - Hook: https://obsigna.dev/hook/overview/ - Dashboard: https://obsigna.dev/dashboard/overview/ - Protocol specification: https://agentreceipts.ai/specification/overview/ - GitHub: https://github.com/agent-receipts/obsigna ## Components - mcp-proxy: MCP STDIO proxy that audits tool calls, applies policy, and emits receipts - hook: PostToolUse hook for Claude Code and other agent runtimes - daemon: Out-of-agent collector that signs and stores receipts - dashboard: Read UI for activity timelines, error rates, and per-tool breakdowns - sdk-go: Go SDK for creating and verifying receipts - sdk-ts: TypeScript SDK (@obsigna/sdk-ts) - sdk-py: Python SDK (obsigna) - openclaw: OpenClaw plugin for automatic receipt generation ## Core concepts - Audit boundary: Receipts are signed outside the agent process, so an agent cannot forge its own audit trail - Agent Receipt: A signed record of a single agent action (W3C Verifiable Credential, Ed25519, SHA-256) - Receipt Chain: Hash-linked sequence of receipts forming a tamper-evident log - Policy: Rules the MCP proxy enforces (allow, deny, require approval) before a tool call runs