Guides
Each one answers a question people ask at the moment it matters: when the bill does not match, when a limit is closing in, when a notification never came. Each is dated and stamped with the version it was tested on, and every figure in it links to the rule behind it.
Cost and limits
Why your Claude Code cost estimate doesn't match your bill
Every rule behind the estimate, its primary sources, and the eight places it is known to differ from what Anthropic bills.
24 September 2026 · Tested on Notchmeter 0.9.24
Will you run out before the window resets?
Pace instead of percentage, a run-out range from your own recent hours, and what Claude's peak hours change.
24 September 2026 · Tested on Notchmeter 0.9.24
Which assistant should get this task?
Switch models before tools, wait when the reset is close, and route by the room each plan has left.
24 September 2026 · Tested on Notchmeter 0.9.24
How much is your prompt cache actually costing you?
What a cache miss costs against a hit, the four causes Claude Code reports, and the tier shift nobody mentions.
24 September 2026 · Tested on Notchmeter 0.9.24
Cursor, Copilot and Antigravity limits in one place
Three vendors, three ideas of what a limit is: what each figure is, where it comes from, and what is never shown as zero.
24 September 2026 · Tested on Notchmeter 0.9.24
Sessions and hooks
Claude Code notifications not working?
Fix Claude Code's own notifications first, then what a hook adds, how to fix one that went quiet, and when not to install it.
24 September 2026 · Tested on Notchmeter 0.9.24
Monitoring coding agents without letting an app write to your config
The meters write nothing; the hook is added only on your click, after a backup. What each tier can know.
24 September 2026 · Tested on Notchmeter 0.9.24
The field
The rules behind every figure
The guides quote it; the accuracy document is where each rule, its source and its date are written down, and where a test pins it.