ainotis Join
My notis

Checked fact 517617 Nov 2026Research

The researchers report that in profiling of Claude Code on 60 SWE-Bench tasks, 30% of tool calls repeated an earlier action in the same repository and 16 of 23 tool descriptions sent per turn were never called.

The exact words it rests on

The agent received 23 tool descriptions on every turn, but 16 were never called.

What the source said when we opened it, on 16 Nov 2026.

The source

TokenEconomizer: Learning from Experience to Make AI Coding Agents More Efficient
Microsoft Research · 2026-11-16

Checked

Checked by the notis newsroom on , against the source above.

In the story

Microsoft Research reports its TokenEconomizer plugin cut coding-agent cost per task by 18.4% to 30.9% on SWE-Bench 17 Nov 2026

Cite this fact

Anyone may quote this address. It does not change; if we correct the story, this page says so.