Token savings in Claude Code and Codex
The base context is re-read on every turn, so a small saving compounds over long conversations. AgentVera lets you choose a context profile per agent, shows the context growing in the pane and compacts at your threshold.
Context profiles
Measurements are the context sent on the first request of a new session (Claude Code 2.1, Codex 0.155).
- Standard: Claude ≈34.4k, Codex ≈13.3k tokens.
- Balanced: only the Bash, Read, Edit, Write, Glob and Grep tools. Claude ≈23.7k tokens (−31%).
- Lean: user plugins, hooks, skills and external MCP are off too. Claude ≈13.3k (−61%), Codex ≈9.7k tokens (−27%).
Compaction and other savings
The pane shows live context size. Run /compact by hand or automatically at a threshold you set. Prompt revision uses a cheaper model, flows can cap the forwarded text, and the Transform node passes on only the part of a long reply you need.
Questions
What does the Lean profile turn off?
It narrows the tool set and doesn’t load user-level plugins, hooks, skills or external MCP. AgentVera’s own MCP servers are still added.
Is there a Balanced profile for Codex?
No. Codex has Standard and Lean.
Are these numbers the same in every project?
No. They were measured on the first request of a new session; your project’s own instruction files and tools change the result.
Related features
Bring your agents to one desk.
Start on the free plan. Download it, add a project folder and open your first agent.
Posts on this topic
Blog- October 4, 2026Hitting Claude Code usage limits? What to doWhen Claude Code usage limits hit mid-task: track remaining usage, shrink context and set up account chains that switch within your provider’s terms.
- September 24, 2026What is an MCP server? Setup for coding agentsWhat is an MCP server and how do you connect one to coding agents? stdio and HTTP servers, token cost, secrets and the right scope, explained step by step.
- September 22, 2026Cut Claude Code token usage by 61%Reduce Claude Code token usage with context profiles, /compact and a narrower tool set: a measured guide that takes the first request from 34.4k to 13.3k.