Guides to tracking API spend
Concrete answers to the questions that come up once your app runs on a dozen metered APIs: where the balance is, how long it lasts, what's burning it, and how to let your coding agent watch it.
Give Claude Code, Cursor or Codex a read-only view of every metered API you build with, so it can check what a run cost, spot a leak, and notice a provider running dry.
What each provider actually exposes over its API, how to put balances, daily spend and plan quotas on one ledger, and where you'll have to estimate.
A runaway loop, a forgotten cron, a retry storm: how to find the call that's draining a balance, from the burn rate down to the endpoint.
What each console's own alerts do, what they can't see, and how to alert on days of runway left instead of a dollar threshold.
The balance endpoint, auth header and response field for every major metered API, and the ones that only show a balance in the console.
Days to zero is balance over burn rate. The hard part is the burn rate: top-ups, daily buckets and credits that aren't dollars.
Daily buckets, async settlement, cache pricing, per-request fees and charges that only bill on a match. Where the numbers drift, and how to reconcile them.
The status code and error each provider returns at zero balance. Several look like rate limits or bad keys, so they get retried instead of fixed.
Which providers send remaining-requests headers, which send credit headers, and which send nothing, so you know before the 429.
Connect a provider key, point Claude Code or Cursor at the Argus MCP server, and ask what the last run cost.