Topic
How platform request quotas are counted, exhausted, and budgeted by clients.
agentgateway v1.6.0 priced a Claude Sonnet 5 request at $0.000156 from its built-in catalog, with no hand-kept modelCatalog in the config. The test covered one Anthropic model, so retiring rate tables is safe only for models the catalog provably knows.
Reality
A developer's signup route issued six API keys overnight to a bot using dot-mutated Gmail addresses. Stripped of dots they were six strangers' inboxes, so canonicalizing emails would not have stopped it.
Reality
Reservation expiry belongs in the database row, a dev.to guide argues, because 300 holds ending at once need five minutes of a 60-per-minute notification API. With the two separated, a throttled notification backlog can grow for minutes without keeping a single item off sale.
Reality
Doubling the worker pool against Nansen's API bought 5.8 seconds of median wall clock across four tokens and cost three calls in 441. Holder Atlas's defaults are back at 4-wide under 5 requests per second.
Reality
A refresh job builds its keep list from successful GitHub responses, so a throttled request looks exactly like a repo that was curated out of the seed. The published fix is a per-slug failure counter that suppresses the prune.
Reality
A developer's Aave liquidation monitor went silent five minutes into a free-tier Alchemy WebSocket. The getLogs loop written to replace it samples ten blocks on a fifteen-minute timer, so most blocks are never read.
Reality
A dev.to walkthrough rebuilds bulk video migration as a SQLite state machine. The arithmetic underneath it: 9,612 assets is roughly 40 minutes of the whole account's API budget.
Reality
A measured run published on dev.to shows three If-None-Match requests returning 304 with empty bodies each took one off the 60/hour bucket. The discount is documented only for authorized calls.
Reality