Module runtime 

Runtime support: key tables, plan cache, federation, runners, and fabric.

Re-exports

pub use cache::OpenAiPlanCache;

pub use cache::PlanCacheKey;

pub use cache::PlanCacheMode;

pub use cache::PlanCacheWriteTarget;

pub use fabric::OpenAiGatewayFabric;

pub use federation::OpenAiFederatedGateway;

pub use federation::OpenAiFederation;

pub use federation::OpenAiFederationPolicy;

pub use keys::OPENAI_GATEWAY_KEY_OBJECT;

pub use keys::OpenAiGatewayKey;

pub use keys::OpenAiKeyTable;

pub use keys::global_openai_key_table;

pub use keys::grant_capability_set;

pub use keys::key_hash;

pub use keys::redacted_gateway_request;

pub use runners::OpenAiRunnerRegistry;

pub use tool_loop::ToolLoopConfig;

pub use tool_loop::run_tool_loop_with_cache;

pub use tool_loop::run_tool_loop_with_registry;

Modules

cache

Plan evaluation cache: keys, modes, write targets, and the in-memory store.

fabric

The OpenAI gateway eval fabric implementing the EvalFabric surface.

federation

Federated inference: remote gateway registry and request/response policy.

keys

API key tables, secret hashing, and per-request capability resolution.

runners

Registry mapping OpenAI model ids to local model runners.

tool_loop

Multi-round tool-call loop driving model runs that invoke tools.