{"slug": "codeburn", "title": "★ CodeBurn", "summary": "CodeBurn, a free, open-source, local-first tool from developer Thiago Perrotta, tracks AI coding token usage and cost across 41 tools and agents including Claude Code, Pi, Codex, and Cursor, broken down by model, project, and task. It reads session files already on disk, requiring no wrapper, proxy, or API key, and offers a TUI, web, desktop, and macOS menubar interface. A 30-day yield analysis showed $594.85 (31%) of $1,915.61 in AI spend was productive, $611.66 (32%) abandoned, and $709.10 (37%) ambiguous across 714 sessions.", "body_md": "♠ [Previously](https://perrotta.dev/2025/09/claude-ccusage/).\n\n[ CodeBurn](https://github.com/getagentseal/codeburn) shows Gen AI / LLM code\nusage: where did all the money\n\nCodeBurn is a free, open-source, local-first tool that tracks AI coding token usage and cost across 41 tools and agents (Claude Code, Pi, Codex, Cursor, and more), broken down by model, project, and task.\n\nIt reads the session files already on disk. No wrapper, proxy, or API key is needed, and nothing leaves the machine (great!).\n\nIt’s a TUI.\n\nInstallation methods:\n\n```\n% brew install codeburn\n```\n\nOR\n\n```\n% npx codeburn\n```\n\nThe same data is also available via web and desktop interfaces. On macOS,\n`menubar`\n\ndownloads the native app to `~/Applications`\n\nand launches it:\n\n```\n% codeburn menubar\nResolving CodeBurn Menubar v0.9.20...\nDownloading CodeBurnMenubar-v0.9.20.zip...\nVerifying checksum...\nUnpacking...\nVerifying app bundle...\nLaunching CodeBurn Menubar...\n\n  Ready. /Users/thiago.perrotta/Applications/CodeBurnMenubar.app\n```\n\nIt shows the current spend in the menu bar. Clicking it opens local breakdowns by agent, model, and activity, plus trends, forecasts, and exports. It refreshes every 30 seconds by default and backs off on battery.\n\nIt tracks spend calculated from local sessions, not provider usage windows or\nreset countdowns. [Codexbar](https://perrotta.dev/2026/05/codexbar/) does the latter.\n\nI find that the CLI TUI is enough.\n\nA few subcommands:\n\n`yield`\n\ncorrelates AI sessions with nearby Git commits. It classifies the money\nspent as productive, reverted, abandoned, or ambiguous:\n\n```\n% codeburn yield -p 30days\n\n  Analyzing yield for Last 30 Days...\n\nProductive:   $594.85 (31%) - 117 sessions shipped to main\nReverted:       $0.00 (0%) - 0 sessions were reverted\nAbandoned:    $611.66 (32%) - 377 sessions never committed\nAmbiguous:    $709.10 (37%) - 220 sessions lost commits to concurrent sessions\n\nAttribution: timestamp-window based (heuristic)\n\nTotal:       $1915.61     - 714 sessions\n```\n\nIt’s an experimental timestamp-based heuristic, but a neat answer to “did all those tokens produce code that shipped?”\n\n`report`\n\nopens the interactive dashboard for a given period, with token and cost\nbreakdowns by tool, model, project, and task:\n\n```\n% codeburn report --provider pi -p month\n```\n\n∎\n\n— § —\n\nReply via [email](mailto:serendipity@perrotta.dev?subject=Reply to: CodeBurn)", "url": "https://wpnews.pro/news/codeburn", "canonical_source": "https://perrotta.dev/2026/08/codeburn/", "published_at": "2026-08-22 22:17:38+00:00", "updated_at": "2026-08-22 22:43:51.280086+00:00", "lang": "en", "topics": ["developer-tools", "ai-tools", "ai-products", "artificial-intelligence"], "entities": ["CodeBurn", "Thiago Perrotta", "Claude Code", "Pi", "Codex", "Cursor", "Codexbar"], "alternates": {"html": "https://wpnews.pro/news/codeburn", "markdown": "https://wpnews.pro/news/codeburn.md", "text": "https://wpnews.pro/news/codeburn.txt", "jsonld": "https://wpnews.pro/news/codeburn.jsonld"}}