The agent provides a ranked list of tools for a specific category, including scores, methodology, and artifact-backed evidence for each ranking decision.
The agent identifies the relevant ranking slug or searches for a topic using list_rankings or search. It then calls get_ranking to retrieve the structured data, including the winner, per-metric scores, and testing methodology. To verify the claims, the agent can optionally call get_evidence to retrieve raw observation cells with screenshots and researcher notes, ensuring the recommendation is backed by provable test data rather than marketing prose.
- Category: Research and knowledge
- Effect: Read-only
- MCP servers: AI Demos MCP (https://koskamo.com/explore/ai-demos-mcp)
- Source: MCP Server — Developer Guide (https://aidemos.com/docs/mcp)
- Source quote: "get_ranking({"slug": "resume-parsing-api"}) returns ranked tools with badges, per-metric scores, the testing methodology, and per-tool worked/struggled breakdowns as typed JSON."
- Playbook page: https://koskamo.com/explore/use-cases/retrieve-best-tool-ranking-with-evidence
Prompt: Find the best OCR tool for invoices and provide the top 3 ranked options with their scores and the evidence behind the ranking.