We asked Claude to review every open PR in the CodeBurn repo.
It spun up 11 review agents. One call alone ran 2m 29s.
That single prompt came to $18.70 at API rates, and $12.98 of it was the helpers.
Do you know what your agents cost per prompt?
@ClaudeDevs Same idea for coding agents. This is what CodeBurn shows for a Claude Code session, locally, no setup: what it cost, what the sub-agents added, and which prompt drove it. Retries are almost always the hidden part.
Introducing CodeBurn for VS Code & Cursor!
Available today. CLI, Desktop, Menu bar, and now Extension.
See today's AI spend in the status bar, open the full dashboard in a tab, find out why a session cost what it did, track cost per pull request, and fix waste with Optimize. 42 AI coding tools, all local, no login, no telemetry.
claude's prompt cache runs on a timer. 5 min by default on the API, 1 hour on claude code. go idle past it and your next message reloads the whole context, about 12x what a warm read costs
same 40 calls:
1 break → $1.51
15 breaks → $5.38
Introducing Anatomy
Describe a machine in one line. Claude brings it to life.
- Interactive isometric wireframes you can rotate, inspect and explore
- WebGL shaders for fire, water, plasma and more
- Real 3D motion, with individual parts that rotate and move
Install it as an
All of this is in CodeBurn 0.9.27, out now. open source, runs on your machine, no proxy, no API keys.
Get it at codeburn.app, or just click Update if you already have it.
What it changes day to day: you stop guessing which work is expensive.
A PR that needed 2 follow-up rounds and 2 review runs shows it. A one-shot PR shows that too. You learn what to hand an agent as is and what to split up first.
CodeBurn now prices every pull request.
Open one and see if it went out in one shot or took follow up rounds, what the review runs cost, how long it took to merge, and how the spend splits into building, review and fixes after review.
All from the session files on your disk.
CodeBurn is now part of Anthropic's Claude Startups program.
CodeBurn shows what your AI coding actually costs. Claude Code, Codex, Cursor, Copilot and 38 more tools, by project, model and session, and why a session got expensive.
Open source. Runs on your machine.
I ran CodeBurn and found tool outputs accounted for 76.8% of context tokens in six of my larger sessions. Digging in, I found very verbose test outputs contributed. I'm going to log full output and keep STDIO minimal. Small token savings can add up. Thank you @TorukMakto1406!
The dollars are an estimate on a subscription, and we label them that way. The breakdown is not: which project, which model, which task, where the agent repeated itself, where a session burned its tokens on retries. That is what your rate limit is made of, and CodeBurn shows it session by session, along with the 5-hour and weekly windows before you hit them.
76 Followers 2K FollowingCurated AI SEO tools + playbooks for small biz. We also build websites, SEO, and managed AI employees. Tested picks, not hype.
41 Followers 463 FollowingCo-founder & CTO of MyTherapyNotes.
Obsessed with local LLMs and autonomous agents. Open-source contributor. MSc of Psychology.
402 Followers 334 FollowingBuilding specialist AI datasets • Fine-tuning & evaluating LLMs • ML student • Glassblower by trade, Sharing my workflows, also Hardware projects
306 Followers 2K FollowingSenior IT System Engineer - Passionate for PowerShell Development - Currently developing advanced functionality on top of Chocolatey for organizations
5K Followers 7K FollowingChief AI Scientist & Director of CibraLAB | CEO of CiberTECCH. Consultant, instructor & international speaker. 🤖 Secure & trustworthy AI systems. 🌐 Catalan.
13 Followers 182 FollowingFull-stack engineer, 15+ years. Building Adlega, an AI CFO for founders. The AI never does the math, a deterministic engine does.
754 Followers 3K FollowingExpert at converting requirements to software!
I build technology that scales—combining deep technical expertise with business goals
Opinions Mine.
5 Followers 140 FollowingTechnical Program Manager in quantum computing | Building practical AI tools | Cloud, engineering delivery & complex systems | PMP · CEng
70 Followers 2 FollowingAntivirus for AI agents. Scan prompts, guard your machine, monitor MCP servers, and detect toxic data flows. 300+ attack probes real-time protection.
136 Followers 2K FollowingWe underuse agents. https://t.co/qciExRvzCv is a free dev tool to improve token use, safety, productivity. Ex Director @PermisoSecurity, CTO @Printivity.
577 Followers 1K FollowingJust a collection of thoughts with a bit of sarcasm about side projects, web development and leadership written by a human.
Ex-Nomad (37+ Countries, 18% 🌍)
67K Followers 105 FollowingWe run the highest signal technical conferences in the industry and publish it all for free online for over 1.5 million AI Engineers a month
15K Followers 2K FollowingCartoonist, Engineer, PM, Partner @a16z investing in infra & AI
Prev Product lead @HashiCorp, Founding Eng/PM @Transposit. Eng @AppDynamics. Opinions = own.
65K Followers 136 FollowingTrusted crypto infrastructure to power your business. We offer a broad suite of products across payments, trading, wallets, and stablecoins.
301 Followers 714 FollowingFounder of @Substratumai | Co-Founder @AetheraiSystems | “If you don’t like the way something is built, build it differently” 🇺🇸