Which tools do Claude, Codex and Cursor choose? We measured 17k runs to find out
What happened
Researchers ran 17,000 experiments to see which external tools (like debuggers, linters, or search engines) AI coding assistants such as Anthropic’s Claude, OpenAI’s Codex, and the Cursor IDE automatically pick when solving programming tasks. The study found each model has distinct preferences—Claude leans heavily on code‑search tools, Codex favors static analysis, while Cursor prefers integrated debugging utilities. Understanding these tool‑selection patterns helps developers predict how AI assistants will behave in real‑world coding workflows.
Why it matters
The findings highlight that the effectiveness of AI coding agents depends not just on the model itself but also on the ecosystem of supporting tools, prompting vendors to optimize tool integration. This could drive a new wave of specialized tooling and tighter AI‑IDE partnerships across the industry.
Is this a big deal?
Get stories like this every morning
The day's AI news in plain English, in your inbox at 8:30 IST. Free.
More from Anthropic
Anthropic Upgrades Claude Code Projects to Run Parallel AI Workers Hands-FreeAnthropic · 14 Nov 2026Claude designed FPGA scaler board for 6th-gen consolesAnthropic · 13 Nov 2026I asked Claude Fable 5.2 to turn my site into a 90s Silicon Graphics workstationAnthropic · 13 Nov 2026Anthropic is endorsing SB 53Anthropic · 13 Nov 2026Understanding and addressing AI harmsAnthropic · 13 Nov 2026This page was collected from armature.tech and summarised automatically by AIPulsee's pipeline. See how it runs.