fix(core): preserve prompt cache affinity

packages/core/src/session/runner/llm.ts

@@ -204,7 +204,10 @@ const layer = Layer.effect(
  const promptCacheKey = /^ses_[0-9a-f]{64}$/.test(session.id) ? session.id.slice(4) : session.id
  const request = LLM.request({
  model,
- providerOptions: { openai: { promptCacheKey } },
+ providerOptions: {
+ openai: { promptCacheKey },
+ openrouter: { promptCacheKey },
+ },
  system: [agent.info?.system, system.baseline]
  .filter((part): part is string => part !== undefined && part.length > 0)
  .map(SystemPart.make),

packages/core/test/session-runner.test.ts

@@ -2507,6 +2507,10 @@ describe("SessionRunnerLLM", () => {
  sessionID,
  otherSessionID,
  ])
+ expect(requests.map((request) => request.providerOptions?.openrouter?.promptCacheKey)).toEqual([
+ sessionID,
+ otherSessionID,
+ ])
  yield* Deferred.succeed(streamGate, undefined)
  yield* Fiber.join(first)
  yield* Fiber.join(second)

packages/llm/src/protocols/openai-chat.ts

@@ -95,6 +95,7 @@ export const bodyFields = {
  tool_choice: Schema.optional(OpenAIChatToolChoice),
  stream: Schema.Literal(true),
  stream_options: Schema.optional(Schema.Struct({ include_usage: Schema.Boolean })),
+ prompt_cache_key: Schema.optional(Schema.String),
  store: Schema.optional(Schema.Boolean),
  reasoning_effort: Schema.optional(OpenAIOptions.OpenAIReasoningEffort),
  max_tokens: Schema.optional(Schema.Number),
@@ -332,11 +333,13 @@ const lowerMessages = Effect.fn("OpenAIChat.lowerMessages")(function* (request:
 
 const lowerOptions = Effect.fn("OpenAIChat.lowerOptions")(function* (request: LLMRequest) {
  const store = OpenAIOptions.store(request)
+ const promptCacheKey = request.model.provider === "openai" ? OpenAIOptions.promptCacheKey(request) : undefined
  const reasoningEffort = OpenAIOptions.reasoningEffort(request)
  if (reasoningEffort && !OpenAIOptions.isReasoningEffort(reasoningEffort))
  return yield* invalid(`OpenAI Chat does not support reasoning effort ${reasoningEffort}`)
  return {
  ...(store !== undefined ? { store } : {}),
+ ...(promptCacheKey ? { prompt_cache_key: promptCacheKey } : {}),
  ...(reasoningEffort ? { reasoning_effort: reasoningEffort } : {}),
  }
 })

packages/llm/test/provider/openai-chat.test.ts

@@ -29,6 +29,20 @@ const request = LLM.request({
 })
 
 describe("OpenAI Chat route", () => {
+ it.effect("maps the prompt cache key for native OpenAI models", () =>
+ Effect.gen(function* () {
+ const prepared = yield* LLMClient.prepare<OpenAIChat.OpenAIChatBody>(
+ LLM.request({
+ model: OpenAI.configure({ baseURL: "https://api.openai.test/v1/", apiKey: "test" }).chat("gpt-4o-mini"),
+ prompt: "hello",
+ providerOptions: { openai: { promptCacheKey: "session_123" } },
+ }),
+ )
+
+ expect(prepared.body.prompt_cache_key).toBe("session_123")
+ }),
+ )
+
  it.effect("prepares OpenAI Chat payload", () =>
  Effect.gen(function* () {
  // Pass the OpenAIChat payload type so `prepared.body` is statically

packages/llm/test/provider/openai-compatible-chat.test.ts

@@ -50,6 +50,18 @@ const providerFamilies = [
 ] as const
 
 describe("OpenAI-compatible Chat route", () => {
+ it.effect("does not send native OpenAI cache options", () =>
+ Effect.gen(function* () {
+ const prepared = yield* LLMClient.prepare(
+ LLM.updateRequest(request, {
+ providerOptions: { openai: { promptCacheKey: "session_123" } },
+ }),
+ )
+
+ expect(prepared.body).not.toHaveProperty("prompt_cache_key")
+ }),
+ )
+
  it.effect("prepares generic Chat target", () =>
  Effect.gen(function* () {
  const prepared = yield* LLMClient.prepare(