Claude Code built recurring AI jobs with no spending cap, burning a user's weekly allowance plus $45.61 in overage
hirobaw · ghdev · 2026-11-16
- A GitHub issue against anthropics/claude-code reports that Claude Code (Opus, desktop app) turned simple condition checks ("new mail?", "is the process running?") into full AI executions every 5-10 minutes on a third-party agent service: 580 runs/day at 40k tokens each.
- No cost estimate, no check of the provider's weekly limit (361.6M tokens), no spending cap, and no brake on repeated failures (one failure loop ran 39 times). The allowance ran out in 3.83 days, followed by $45.61 in paid overage. Claude itself conceded in-session: "The USD 40 is money lost because of my design."
- Support fared no better: automated replies, an unrelated "service disruption" response, then a final refusal to assess the link between Claude Code's design and the loss. Anthropic's Gmail connector also refuses to reply in threads containing Anthropic's own automated emails, cutting off the support conversation.
- The user asks for cost/limit estimation and spending safeguards for scheduled jobs, plus a working support escalation path.
More from coding & agent
- llms.txt Protocol Goes Mainstream as Google Antigravity Adopts It Two Years On — jeremyphoward · 2026-11-22
- He built a feed only agents can read: no UI, semantic matching, agent-rated posts — imdogefather · 2026-11-22
- Claude Code Debuts Experimental Code Mode as AI-Driven Tooling Gains Momentum — lxfater · 2026-11-22
- ClaudeAI Weekly: Single-HTML-File Arms Race, 8.9B-Token Strawberry, Post-IPO Anthropic — ClaudeAI-mod-bot · 2026-11-22
- Agno's AgentOS turns any agent into a production FastAPI app with MCP, RBAC and tracing — pritisinghhhh · 2026-11-22
- Chinese devs vent: app store rejections and copycats are just the norm — tinyfool · 2026-11-22