Module prompt_cache 

Structs

AnthropicPromptCacheSettings

Anthropic Claude cache control settings

DeepSeekPromptCacheSettings

DeepSeek prompt caching configuration (automatic KV cache reuse)

GeminiPromptCacheSettings

Gemini API caching preferences

MoonshotPromptCacheSettings

Moonshot prompt caching configuration (leverages server-side reuse)

OpenAIPromptCacheSettings

OpenAI prompt caching controls (automatic with metrics)

OpenRouterPromptCacheSettings

OpenRouter passthrough caching controls

PromptCachingConfig

Global prompt caching configuration loaded from vtcode.toml

ProviderPromptCachingConfig

Per-provider configuration overrides

ZaiPromptCacheSettings

Z.AI prompt caching configuration (disabled until platform exposes metrics)

Enums

GeminiPromptCacheMode

Gemini prompt caching mode selection

OpenAIPromptCacheKeyMode

OpenAI prompt cache key derivation mode.

PromptCacheRetention

OpenAI prompt cache retention policy. Controls the model-side prompt caching window.

Functions

build_openai_prompt_cache_key

Build a stable OpenAI prompt_cache_key for requests that should share provider-side cache routing.

build_session_affinity_prompt_cache_key

Build the namespaced session-affinity / prompt-cache key for a request.

map_prompt_cache_key_for_provider

Rewrite an OpenAI-style cache key for routes that share the wire shape but need distinct routing stickiness. Keeps the stable session identifier while namespacing gateway/session-affinity traffic apart from native OpenAI.

session_affinity_key_enabled

Whether LLMRequest.prompt_cache_key should carry session lineage for this provider.

session_affinity_provider

Providers that send lineage-stable session/cache identity on the wire.