Glossary

21 terms a developer or an agent operator runs into.

Context window — The maximum number of tokens a model can consider at once — prompt and response together.
Token — The unit a language model reads and bills in — roughly a word fragment, not a character or a word.
Prompt caching — Reusing the model's processed form of a repeated prefix so it is not paid for or recomputed on every request.
Tool call — A structured request from a model to run a function you defined, with arguments it filled in.
MCP (Model Context Protocol) — An open protocol for exposing tools, data and prompts to an AI client through one standard interface.
Idempotency key — A client-supplied identifier that lets a request be retried safely without repeating its effect. (full guide)
Exponential backoff — Retrying a failed request after a delay that doubles each attempt, so a struggling server is not hammered. (full guide)
Rate limit — A cap on how many requests a client may make in a window, enforced by the server. (full guide)
HTTP 429 Too Many Requests — The status a server returns when a client has exceeded its rate limit. (full guide)
ETag — An opaque version identifier for a resource, used for caching and for detecting concurrent writes. (full guide)
Optimistic locking — Allowing concurrent writes and rejecting the ones that would overwrite a change made since the writer last read. (full guide)
Cursor pagination — Paging through a list with an opaque pointer to the last item seen, instead of a numeric offset. (full guide)
At-least-once delivery — A guarantee that a message will arrive, with no promise it arrives only once. (full guide)
Webhook signature — A cryptographic header proving a webhook came from the sender you expect and was not altered. (full guide)
Backpressure — A slow consumer signalling upstream to slow down, instead of silently accumulating work. (full guide)
GEO (Generative Engine Optimization) — Making a page usable as a source by AI answer engines, rather than only rankable by search engines.
llms.txt — A proposed file at a site's root that points AI systems at its most useful documentation.
robots.txt — A root file telling crawlers which paths they may fetch — advisory, not enforcement.
Server-side rendering (SSR) — Producing a page's HTML on the server so it is readable before any JavaScript runs.
Cron expression — A five-field schedule string: minute, hour, day of month, month, day of week. (full guide)
TTL (time to live) — How long a cached or stored value stays valid before it must be refreshed or discarded. (full guide)