Compare two tools head-to-head
The agent delivers a provable comparison between two specific tools, highlighting differences in performance on identical test inputs with artifact-backed evidence.
How it works
Read-onlyResearch and knowledge
The agent identifies the slugs of the two tools to compare. It then calls compare_tools to retrieve a head-to-head analysis based on the same input scenarios. This ensures the comparison is structurally honest and not based on disparate testing conditions. The agent reviews the verdicts, notes, and artifacts (screenshots) to determine which tool performs better for the specific use case.
Prompt
Paste this into an agent that has the servers below connected. Adjust the names and dates to your own.
Prompt for Compare two tools head-to-head
Compare Tool A and Tool B for invoice processing using the compare_tools function and show the evidence for their performance differences.MCP servers it needs
Source
The vendor page this playbook is drawn from, and the passage that supports it.
Two cells on the same scenario = the tools saw the same input — that is what makes compare_tools.head_to_head a provable comparison rather than marketing prose.MCP Server — Developer Guide