arsentev-ai/contextburn

Overview

Local Python tool that measures the share of paid tokens your coding agent converts into output versus context re-reading. Runs entirely on your machine with no network calls, giving you a direct view of token spend efficiency.

Best for

Best for
Developers who want to quantify coding agent token waste

Use cases

• Audit token consumption of coding agent sessions
• Identify how much context re-reading costs
• Track output efficiency across prompts

How to use

Install

claude mcp add contextburn -- uvx contextburn mcp

Tools exposed

• CONTEXTBURN_DAY_START
• CONTEXTBURN_WARN
• CONTEXTBURN_ALARM

Tested with

Claude Code, Cursor, Windsurf, VS Code

Notes

Local Python tool that measures the share of paid tokens your coding agent converts into output versus context re-reading. Runs entirely on your machine with no network calls, giving you a direct view of token spend efficiency.

0 stars on GitHub. Last updated 2026-09-13. Licensed MIT.

Use cases

• Audit token consumption of coding agent sessions
• Identify how much context re-reading costs
• Track output efficiency across prompts

Pros

• Fully local, no network calls
• Direct measurement of work vs context overhead
• Simple Python tool

Cons

• No stars or community validation yet
• Requires Python environment
• Only measures token share, not quality of output

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.