---
title: "Research: how AI assistants recommend tools, measured"
canonical_url: https://cookiesforai.app/research
last_updated: 2026-10-03
type: index
summary: "Our own dated tests of how ChatGPT, Gemini, Perplexity, Google AI Mode, Grok and Claude answer questions about software, agent skills and MCP servers, with the raw data."
---

# Research

Our own dated tests of how ChatGPT, Gemini, Perplexity, Google AI Mode, Grok and Claude answer questions about software, agent skills and MCP servers, with the raw data.

- [Agent skill and AI tool directories compared, October 2026](https://cookiesforai.app/research/agent-skill-and-ai-tool-directories-compared-october-2026): A dated comparison of 9 agent skill directories and 6 AI tool directories that AI assistants used in our October 2026 tests: size, item pages, how to get listed, agent extras and how often each was cited. (updated 2026-10-03)
- [How ChatGPT, Gemini, Perplexity, Google and Grok answer software and AI tool questions](https://cookiesforai.app/research/ai-assistants-field-study-october-2026): We asked ChatGPT, Gemini, Perplexity, Google AI Mode, Google AI Overviews and Grok the same 13 questions about software and AI tools, 110 times on 2026-10-01, and recorded every name and source they gave. (updated 2026-10-01)
- [How Claude finds agent skills it does not know](https://cookiesforai.app/research/how-claude-finds-agent-skills-october-2026): In 93 runs on Claude Opus 5.5, Sonnet 5.5 and Fable 5.1 on 2026-10-01, every question about a little-known agent skill made Claude search, and 13 of 18 answers linked a skill directory page. (updated 2026-10-01)
- [AI assistants: model, search and crawler changes, dated](https://cookiesforai.app/research/engine-changes): The engine changes tracker lists dated changes to the models, web search and crawlers of Claude, ChatGPT, Google (Gemini, Search and AI features), Grok, Perplexity, Copilot and open models, one table per assistant, each row linked to the maker's own page. (updated 2026-10-03)