Agent Receipts

Your AI agent remembers everything — and you can prove it.

Persistent memory with cryptographic receipts. Every fact your agent learns is signed, traceable, and verifiable. Local-first, zero config, free forever.

Want to try it first? Run the interactive demo →

Step 1 — Add to your MCP config

{
  "mcpServers": {
    "agent-receipts": {
      "command": "npx",
      "args": ["@agent-receipts/mcp-server"]
    }
  }
}

Config locations: Claude Desktop: ~/Library/Application Support/Claude/claude_desktop_config.json · Claude Code: .mcp.json · Cursor: .cursor/mcp.json

Step 2 — Restart your AI client

Restart Claude Desktop, Cursor, or VS Code. Agent Receipts will appear in your available tools.

Step 3 — Start an agent task

Every action your agent takes now generates a signed receipt. View them in the local dashboard:

npx @agent-receipts/dashboard

GitHub →How it works View demo dashboard

@agent-receipts/mcp-server · MIT License