CodeBurn: Track Your AI Spend Across 32 Tools from the Terminal

You’re building with AI tools like ChatGPT, Claude, GitHub Copilot, and a dozen others. But when the invoice comes, it’s a mess. Each platform bills differently, and you have no single view of your total AI spend.

CodeBurn is a local CLI that solves exactly that. It scans your usage across 32 AI tools and maps your spending into one clean terminal report. No cloud uploads, no sign-ups. Just codeburn scan and you know where your money went.

What It Does

CodeBurn is a Go-based command-line tool that aggregates cost data from popular AI services. It works by reading local configuration files, API keys, and usage logs stored on your machine (or fetched via public APIs you authorize). Currently it supports 32 tools including:

• OpenAI (ChatGPT, API usage)
• Anthropic (Claude)
• GitHub Copilot
• Replicate
• ElevenLabs
• Midjourney (via Discord logs)
• And 26 others

Run codeburn scan and it outputs a table showing per-tool spend, monthly totals, and trends. You can filter by date range or tool category.

Why It’s Cool

Three things stand out:

First, it’s fully local. Your API keys and usage data never leave your machine. No SaaS, no cloud backend. Just a binary you run in your terminal. This matters for anyone working with sensitive data or under compliance requirements.

Second, it’s surprisingly broad. 32 tools is a lot. Most spend trackers only cover ChatGPT and Copilot. CodeBurn reaches into less obvious ones like Stable Diffusion, Replicate, and even niche APIs. If you’re a power user, you’ll likely find your full stack represented.

Third, the implementation is practical. Instead of scraping or requiring custom webhooks, it uses standard sources: config files (like ~/.copilot.json), environment variables, and CLI history. This means setup is quick, and you’re not locked into a specific workflow.

How to Try It

Assuming you have Go installed (1.21+), you can install CodeBurn in two commands:

go install github.com/AgentSeal/codeburn@latest
codeburn scan

For the full list of supported tools and their setup, check the project’s README:
github.com/AgentSeal/codeburn

If you’re on macOS, there’s also a Homebrew tap in the repo for easier updates.

Final Thoughts

I’ve been using this for the past week, and it’s genuinely useful. Before CodeBurn, I had no idea how much I was spending on ElevenLabs vs. Copilot. Turns out, my voice cloning habit was costing more than my code generation. Who knew.

If you’re a solo dev, freelancer, or small team running multiple AI tools, give it a try. It’s lightweight, open source, and does one thing well. Plus, it’s nice to finally have a single dashboard for AI costs that doesn’t require selling your data to a third party.

Found this useful? Follow @githubprojects for more developer tools and open source projects.