What Is an MCP Execution Receipt
MCP execution receipt definition
An MCP execution receipt is a public page that records one MCP tool call: which server ran, which tool was invoked, with what arguments, what was observed, and a fingerprint binding the record to that outcome.
To verify an MCP tool call, open the receipt URL, check the server and tool name, and recompute the SHA-256 fingerprint against the observed result.
Every MCP tool call on Zambo mints an execution receipt, so any agent action is independently checkable by anyone with the link.
MCP is the connection protocol between an AI client and a tool server. The receipt is the public record of the call that crossed that boundary. It lets a reviewer inspect the server, tool, arguments, observed result, and integrity fingerprint without relying on a chat summary.
Live receipt examples
live_price receiptOpen the public record and inspect its verification path. capability_search receiptOpen a second public record from the hosted MCP reference implementation.What an MCP receipt records
The server and named capability identify which MCP boundary ran.
The recorded arguments show what the client asked the tool to do.
The receipt reports what the implementation returned, without claiming more than it observed.
SHA-256 over the observed result gives reviewers a reproducible integrity check.
How to verify an MCP tool call
- Open the public receipt URL and confirm the receipt ID.
- Check the server name, tool name, and arguments.
- Read the observed result and its status.
- Recompute the SHA-256 fingerprint over the stated representation.
- Report only what the receipt and its evidence support.
Related definitions
Read the broader AI agent execution receipt definition, the AI agent audit trail definition, and the verifiable AI tool calls guide.
The hosted reference connection is documented here. Connect one server at https://zambo.dev/api/mcp.