Claude Code startup context cut from 63K to 22K tokens, saving 65%
dSebastien · x · 2026-11-03
Developer dSebastien reports that after optimizing his vault setup, a fresh Claude Code session starts at about 22K tokens instead of 63K — 65% less context burned before typing a word, directly improving usable window and cost efficiency for heavy users.
Related event: On-Demand Skill Loading Cuts Claude Code Startup Context by 65%(2 posts)→
More from coding & agent
- Virtual Human: formalizing the body as executable graphs so agents can solve biology with code — Vjeux · 2026-11-03
- State of Context Engineering 2026: 6 failure modes and 6 fixes — dbreunig · 2026-11-03
- AI broke Joel Spolsky's rewrite rule: Microsoft's Toub rewrote Copilot's 400k-line runtime in Rust alone — msharmas · 2026-11-03
- Flock: A New iOS App for Running AI Coding Agents, Built on a t3code Fork — seatedro · 2026-11-03
- WhatIf: 10,000+ steerable LLM agents form a flight simulator for policy — scyrusk · 2026-11-03
- LLMs solving math gets all the hype, but their new infrastructure patterns for reliable software go unnoticed — StewartalsopIII · 2026-11-03