TypeSafe AI typesafe¶
The one provider here that generates no text. TypeSafe AI's System One model, Jev, answers a state and a set of typed questions with typed answers, so its routes — typesafe/jev-latest and typesafe/jev-preview — serve decisions and nothing else, over the typesafe-systemone wire, paid by the deployment's own API key.
What it serves¶
Only POST /v1/decisions, reached through decision-model or best-decision-model. The descriptor carries no chat path at all, so a chat completion naming one of its routes is refused before any request is built — invalid_request: route `typesafe/jev-latest` serves typed decisions only; call POST /v1/decisions — rather than sent to an endpoint that does not exist. An operator who wants a decision answered by a chat model points the alias at that model instead: Brama renders the same questions onto it and computes the typed answer itself.
How we authenticate¶
There is no login flow here. TypeSafe AI issues an API key in its own console; an operator stores it once in the Skarbiec vault, and at final use the gateway redeems the deployment's capability for this provider through the local entitlements router, reduces the stored document to the bare key, and sends it as Authorization: Bearer <key>. Until that capability exists on a host, an alias routed here reports capability_absent and names the provider it is waiting for.
The two failure shapes point at two different repairs: a 503 credential_unauthorized means no capability, read grant or trust material produced a key at final use — repair the authorization chain (runbook); a 502 provider_failure with a provider_authentication: prefix means the key was produced and TypeSafe AI rejected it — repair the provider account.
Models¶
Two pinned catalog rows, widened (never replaced) by live discovery through the models endpoint: typesafe/jev-latest and typesafe/jev-preview.
Plan usage¶
TypeSafe AI publishes no free plan-usage report for API keys. The ledger records that absence as the provider's own answer, never as a zero (subscription).
Endpoints¶
| Decisions | https://api.typesafe.ai/v1/systemone |
| Chat | none — this provider serves no chat |
| Model discovery | https://api.typesafe.ai/v1/models |
| Auth header | Authorization: Bearer <key> |
| Trusted host | api.typesafe.ai |
| Base URL override | BRAMA_PROVIDER_TYPESAFE_BASE_URL, validated with the refusals in provider |