Agents & Inference
Claude Code burns 15-20% of tokens on background agents' 3-5 word status lines
Which summary reads better? Pick one — models revealed after.Both summaries are AI-generated.
Summary A
Claude Code sends every background agent's entire transcript to its model every 30 seconds, consuming 15-20% of tokens on a 3-word status update that doesn't affect the conversation. Disabling this feature can be done via a Bun preload script, saving token capacity and rate-limit headroom for production LLM deployments. This tweak enables more efficient use of LLM resources for actual conversations.
Summary B
Mistral Large quota or rate limit — check usage and plan. Original headline: Claude Code regularly sends each background agent’s transcript to its model
0 picks
source & further reading
snipvote.com — original article
Gemini 4 Carbon
Google’s Genie 4 world model adds audio, more realistic worlds, and real-time steering
Skill Dreaming builds repository-specific agent skills from pull request streams