A frustrated developer whose Anthropic Claude quota evaporated in roughly 10 minutes has. Shipped an open-source tool. That dissects exactly how tokens get consumed, exposing the request patterns. That can silently drain usage limits on AI coding assistants. The project, posted to GitHub and discussed on Hacker News, has sparked debate. About transparency in AI subscription metering. This guide covers Show HN: My Claude quota ran out in 10 minutes. So I made a tool to find out why in detail. This guide covers Show HN: My Claude quota ran out in 10 minutes. So I made a tool to find out why in detail. This guide covers Show HN: My Claude quota ran out in 10 minutes, so I made a tool to find out why in detail. This guide covers Show HN: My Claude quota ran out in 10 minutes, so I made a tool to find out why in detail. This guide covers Show HN: My Claude quota ran out in 10 minutes, so I made a tool to find out why in detail.
Show HN: My Claude quota ran out in 10 minutes. So I made a tool to find out why: Show HN: My Claude quota ran out in 10 minutes. So I made a tool to find out why: Show HN: My Claude quota ran out in 10 minutes. So I made a tool to find out why: What happened. When the Claude quota died in 10 minutes?
The creator, who goes by the handle behind the GitHub repository "claude-usage-tracker," reported. That during what they described as a normal coding session, their Claude usage. Allowance was exhausted in approximately 10 minutes, a consumption rate. That seemed implausible for the volume of visible conversation. The tool they subsequently built logs every API call, counts input and output. Tokens, attributes cost per request. And projects remaining budget against hard session limits, according to the project's README.
How does the token-tracking tool actually work?
The utility installs as a lightweight middleware layer. That intercepts requests between a user's coding environment and the Claude API, recording. Token counts returned in Anthropic's response headers. It then generates a local dashboard displaying cumulative spend, average tokens per turn. And a real-time estimate of how long the current quota will last at present usage. Because all processing happens client-side, no conversation data is exfiltrated, a design choice. The author emphasizes in the documentation.
For developers, the practical workflow is straightforward: run the tracker alongside the Claude. CLI or API client, watch the dashboard during long sessions. And trigger an alert when projected burn rate exceeds the remaining window. The tool also flags unusually large system prompts, a common hidden drain. When projects accumulate large instruction files or entire codebases in context.
Why are quotas evaporating so fast, and what is Anthropic actually charging for?
According to Anthropic's published pricing page, Claude Sonnet 4.5 is billed at $3 per million input tokens and $15 per million output tokens, while Claude Opus 4.1 costs $15 and $75 per million tokens respectively. Subscription plans such as Claude Pro at $20 per month and Claude Max. At $200 per month include fixed message allowances. That vary by model, but the fine-grained token math is not surfaced inside. The chat interface itself, meaning users often discover burn rates only. When the session is already cut off.
What This Means
For individual developers and small teams, the episode highlights a structural information gap:. Consumers of frontier AI products rarely see the per-request cost telemetry. That would let them budget effectively. Open-source tooling that exposes this layer effectively functions as a market-correction mechanism, pressuring. Vendors to either publish clearer usage dashboards. Or risk losing power users to more transparent competitors. It also reinforces a broader pattern in. Which developer communities build accountability layers around opaque SaaS pricing, much as they. Once built cost dashboards for AWS and GCP.
What's Next
Anthropic has not publicly responded to the GitHub project, and the tracker remains community-maintained. Expect similar tools to emerge for OpenAI, Google Gemini. And Mistral as usage-based AI pricing spreads, Whether you are new to Show. HN: My Claude quota ran out in 10 minutes. So I made a tool to find out why or already experienced, the. Sections below have you covered. Whether you are new to Show HN: My Claude quota ran out in 10 minutes. So I made a tool to find out why or already experienced, the sections below have you covered. Whether you are new to Show HN: My Claude quota ran out in 10 minutes. So I made a tool to find out why or already experienced, the sections below have you covered. and watch for vendors to begin shipping native quota dashboards before third-party trackers force their hand.
Key Points
GitHub tool "claude-usage-tracker" exposes hidden token consumption in Claude coding sessions
Developer reported a 10-minute quota wipeout during routine work
Anthropic's Claude Sonnet 4.5 is priced at $3 input and $15 output per million tokens
Claude Pro costs $20 monthly, Claude Max costs $200, each with model-dependent message caps
The Bottom Line
If you are burning through Claude quotas faster than your screen time suggests. The gap is almost certainly invisible system prompts, large file contexts, or tool-calling loops. And a lightweight client-side tracker is now the fastest way to prove it.
Related Resources
For more context, check our related article on London's Ultra Low Emission Zone Sparks 'Stunning' Lung Recovery in Children, Study Finds, related article on AWS's 'Quarantine' Policy Leaves Hundreds of Leaked Root Keys Dangerously Active, related article on Beyond Gmail: Why True End-to-End Encrypted Email is Essential for Your Digital Privacy.
