getagentseal/codeburn

What it solves

CodeBurn provides detailed visibility into AI coding spend. While standard billing statements only provide monthly totals, CodeBurn breaks down costs by project, model, tool, and specific tasks (such as coding, debugging, or planning) to identify where tokens are being wasted.

How it works

Instead of requiring an account or sign-up, CodeBurn reads the session files that AI coding tools (like Claude Code, Cursor, and Codex) write to your local disk. It then prices every token and produces a cost breakdown. It offers multiple interfaces, including a terminal dashboard, a desktop app, a macOS menu bar popover, and a browser-based view.

Who it’s for

Developers who use multiple AI coding agents and want to track their actual spending, monitor subscription quotas, and optimize their token usage to reduce costs.

Highlights

• Multi-tool Support: Integrates with 42 different AI coding tools and gateways (e.g., Cursor, Claude Code, GitHub Copilot).
• Cost Optimization: The optimize command identifies token-wasting configurations (like oversized CLAUDE.md files) and can automatically apply fixes.
• Quota Management: Tracks live limits and subscription plans to warn users when they are on pace to exceed their budget.
• Capping Guard: An optional local hook for Claude Code that can warn or stop sessions once specific spending thresholds are reached.
• Agent Integration: Includes an MCP server that allows AI agents to answer questions about your spending and optimization directly within a conversation.