Futures
Access hundreds of perpetual contracts
TradFi
Gold
One platform for global traditional assets
Options
Hot
Trade European-style vanilla options
Unified Account
Maximize your capital efficiency
Demo Trading
Introduction to Futures Trading
Learn the basics of futures trading
Futures Events
Join events to earn rewards
Demo Trading
Use virtual funds to practice risk-free trading
Launch
CandyDrop
Collect candies to earn airdrops
Launchpool
Quick staking, earn potential new tokens
HODLer Airdrop
Hold GT and get massive airdrops for free
Pre-IPOs
Unlock full access to global stock IPOs
Alpha Points
Trade on-chain assets and earn airdrops
Futures Points
Earn futures points and claim airdrop rewards
Promotions
AI
Gate AI
Your all-in-one conversational AI partner
Gate AI Bot
Use Gate AI directly in your social App
GateClaw
Gate Blue Lobster, ready to go
Gate for AI Agent
AI infrastructure, Gate MCP, Skills, and CLI
Gate Skills Hub
10K+ Skills
From office tasks to trading, the all-in-one skill hub makes AI even more useful.
GateRouter
Smartly choose from 40+ AI models, with 0% extra fees
Anthropic billing bug overcharged $200, initially refused to refund, then compensated.
According to Beating Monitoring, there is a billing bug in Claude Code: as long as the commit history of the git repository contains the uppercase string HERMES.md, all API requests will bypass the Max subscription quota and be charged directly from extra usage. One Max 20x user (monthly fee $200) was quietly charged an additional $200.98 for extra usage, even though they only used 13% of their subscription quota.
The trigger condition is extremely precise: it must be the uppercase HERMES.md, case-sensitive, with the .md suffix. lowercase hermes.md is fine, removing the suffix from HERMES is fine, HERMES.txt is also fine. The trigger point is not whether the file exists on disk, but whether Claude Code inserts the git log commit messages into the system prompt, and the server-side anti-abuse system performs string matching on this text. HERMES is the configuration file name for Hermes Agent, an open-source AI agent framework by Nous Research. Anthropic’s anti-abuse system is most likely detecting whether users are calling the API through unofficial clients, but it mistakenly penalized all normal users who mention this filename in their commit messages.
The reporter used binary search for hours to locate the cause: first cloning the repository to test isolated branches, then narrowing down the commit range step by step, finally pinpointing the trigger string. Another user independently reproduced the issue and wrote an automated troubleshooting script. Boris Cherny, head of Claude Code, confirmed it was an “overactive anti-abuse system” and fixed it on the same day.
However, the refund process went awry. After the user submitted a refund request, Anthropic’s AI customer service replied that “compensation cannot be provided for downgraded services or billing routing technical errors,” and directly refused. This reply sparked outrage on GitHub and Hacker News, making the front page of HN. The community’s anger was not just about the $200: the anti-abuse system should not mistake false positives for “quota exhausted” errors, and it certainly shouldn’t refuse refunds after admitting to a bug. Thariq, a member of the Claude Code team, later announced that all affected users would receive a full refund plus equivalent points compensation.