Per-subagent token spend tracking with hard budget caps for Claude Code power users

A heavy Claude Code user burned 80% of a week's quota in 10 minutes because subagents do not report token usage, and their recursive fact-checking loop kept silently running. There is no tool that attributes token spend to each agent and enforces a hard stop. A lightweight companion that reads session logs, shows the live per-agent cost breakdown, and caps spend would be worth paying for, because the user was already deciding between expensive overage tokens and a 20x upgrade.

Signal strength

Why this is worth a look — three verifiable signals, not a black-box score.

• Cross-source

  1 independent platforms · 1 related discussions

  Weak

• Discussion heat

  not yet measured

  Weak

• Freshness

  11 days ago

  Strong

Related topics