Skip to main content

​
Agent

A configured client that sends requests through Manifest. Each agent has its own API key (mnfst_...), its own routing rules, and its own usage page. An agent typically corresponds to one tool or workflow (your IDE plugin, a Slack bot, a scheduled job), not one user.

​
Auth type

The credential category Manifest uses to talk to a provider:
  • api_key — classic per-token API key (most providers)
  • subscription — OAuth or a subscription token tied to a paid plan (ChatGPT Plus, Claude Max, GLM Coding Plan, and others)
  • local — no credential, the server runs on your machine (Ollama, LM Studio, llama.cpp)
Auth type is recorded on every request and shows up in the dashboard’s distribution chart.

​
Auto

The model ID you send to opt into routing. When Manifest sees auto, it applies your routing rules and picks the real model. Send a real model ID instead to skip routing and target one provider directly (see Direct routing).

​
Custom tier

A routing rule matched on a request header. You choose the header key and value on the dashboard and pin the tier to its own model and fallback chain. When a request carries that header, it routes to the tier’s model instead of the default. See Routing → Custom.

​
Direct routing

Sending a real model ID (one that GET /v1/models lists) instead of auto. Manifest forwards the request straight to that provider and skips tiers and fallbacks. The response carries X-Manifest-Tier: direct.

​
Fallback

The retry mechanism that kicks in when the primary model fails. Manifest tries the next model in the tier’s fallback list, then the next, until one succeeds or the list is exhausted. See Fallback for triggers and config.

​
Fallback chain

The ordered list of models tried for a single tier, primary first. Up to 5 models. When all of them fail, Manifest returns HTTP 424 with X-Manifest-Fallback-Exhausted: true.

​
Provider

An upstream LLM service Manifest can route to. There are four kinds:
  • API key — pay-per-token (OpenAI, Anthropic, Google, and more)
  • Subscription — a plan you already pay for (ChatGPT Plus, Claude Max, and more)
  • Local — runs on your hardware (Ollama, LM Studio, llama.cpp)
  • Custom — any OpenAI- or Anthropic-compatible HTTP endpoint

​
Routing

The step that decides which model handles a request. Manifest checks your custom tiers first, then sends anything that doesn’t match to your default model. It runs in-process with no extra network call. See Routing.

​
Tier

A routing lane with its own model and fallback chain. Every agent has a Default tier; you can add Custom tiers matched on request headers. See Routing.