Today for AI
HOT RADAR
toolsHEAT 12.1°

AI CURATED HOT STORY · 10/29/2026

Claude Code Burns 15-20% Tokens on Status Line; Community Patch Released

1 reports archived1 independent sourcesupdated 10/29/2026, 11:33:42
Synthesis & Latest Updates
1 Sources Cross-Validated

Claude Code background agents send full transcripts to the model every 30 seconds to generate brief status descriptions, wasting ~15-20% of tokens. Developers discovered this via OpenTelemetry data and released a preload script to disable it, noting the behavior has been default since version 2.1.232 with no official toggle.

LATEST/Claude Code background agents send full transcripts to the model every 30 seconds to generate brief status descriptions, wasting ~15-20% of tokens. Developers discovered this via OpenTelemetry data and released a preload script to disable it, noting the behavior has been default since version 2.1.232 with no official toggle.

Hourly heat curve

10 fully observed hours
Now
12.1
Peak
15.012:00
24h change
–

The line compares only sources observed throughout; gap hours are interpolated to keep the trend continuous.

Coverage timeline

Total 1 reports · Latest first