How to make Claude Code read 1M tokens while using only 137k of context
umeshonai · x · 2026-10-20
umeshonai shares a 3-step trick to make Claude Code read and understand millions of tokens while using only 137k in the main context: 1) create a .claude/agents folder in the project root; 2) add agent definition files with per-agent model selection and effort level; 3) route the main chat to these agents directly, via CLAUDE.md or AGENTS.md. Subagents hold their own context windows and process documents in parallel, returning only results—screenshots show the main thread at 14% context (137.4k/1M) while background tasks consumed 778k tokens. An included adversarial-reviewer agent definition runs a two-phase pipeline (assumption deconstruction + vulnerability mapping) and outputs risk-ranked XML findings.
More from coding & agent
- Decisions API hits GA as agents automate cloud video editing via feedback loops — dkundel · 2026-10-23
- Swarms API changelog: 8 months of streaming, multi-agent orchestration, and dynamic model pricing — KyeGomezB · 2026-10-23
- A curated library of 50+ free AI agent resources: videos, repos, one Google Doc — ifioknkem · 2026-10-23
- Vercel adds no-account deploys: drag and drop a file to get a live URL — steren · 2026-10-23
- Dev runs 5 Claude Code sessions that coordinate via MCP Agent Mail to root his network cameras — doodlestein · 2026-10-23
- Vercel ships 'vercel deploy --temporary' for ephemeral agent-built deployments — cramforce · 2026-10-23