feat(ai): add OpenCode Go provider

packages/app-expo/src/components/onboarding/steps/AIPage.tsx

@@ -37,6 +37,7 @@ const PROVIDER_OPTIONS: { id: AIProviderType; name: string }[] = [
   { id: "lmstudio", name: "LM Studio" },
   { id: "openrouter", name: "OpenRouter" },
   { id: "siliconflow", name: "SiliconFlow" },
+  { id: "opencode", name: "OpenCode Go" },
   { id: "custom", name: "Custom" },
 ];
 

packages/app-expo/src/screens/settings/ai/EndpointEditor.tsx

@@ -36,6 +36,7 @@ const PROVIDERS: { id: AIProviderType; label: string }[] = [
   { id: "moonshot", label: "Moonshot" },
   { id: "zhipu", label: "智谱 GLM" },
   { id: "aliyun", label: "阿里云通义" },
+  { id: "opencode", label: "OpenCode Go" },
   { id: "custom", label: "Custom" },
 ];
 

packages/app/src/components/onboarding/steps/AIPage.tsx

@@ -29,6 +29,7 @@ const PROVIDER_OPTIONS: { id: AIProviderType; name: string }[] = [
   { id: "lmstudio", name: "LM Studio" },
   { id: "openrouter", name: "OpenRouter" },
   { id: "siliconflow", name: "SiliconFlow" },
+  { id: "opencode", name: "OpenCode Go" },
   { id: "custom", name: "Custom" },
 ];
 

packages/app/src/components/settings/AISettings.tsx

@@ -45,6 +45,7 @@ function useProviderOptions(): { value: AIProviderType; label: string }[] {
     { value: "moonshot", label: "Moonshot (Kimi)" },
     { value: "zhipu", label: t("settings.ai_provider_zhipu") },
     { value: "aliyun", label: t("settings.ai_provider_aliyun") },
+    { value: "opencode", label: "OpenCode Go" },
     { value: "custom", label: t("settings.ai_provider_custom") },
   ], [t]);
 }

packages/core/src/ai/__tests__/llm-provider.test.ts

@@ -142,6 +142,59 @@ describe("getEndpointFetch Gemini thought signatures", () => {
   });
 });
 
+describe("getEndpointFetch OpenCode Go headers", () => {
+  it("identifies the client and forwards the conversation session id", async () => {
+    let capturedHeaders: Headers | undefined;
+    globalThis.fetch = vi.fn(async (input: RequestInfo | URL, init?: RequestInit) => {
+      capturedHeaders = input instanceof Request ? input.headers : new Headers(init?.headers);
+      return new Response("{}", {
+        status: 200,
+        headers: { "content-type": "application/json" },
+      });
+    }) as typeof fetch;
+
+    const endpoint = makeEndpoint({
+      provider: "opencode",
+      baseUrl: "https://opencode.ai/zen/go/v1/chat/completions",
+      useExactRequestUrl: false,
+    });
+    const endpointFetch = getEndpointFetch(endpoint, "some-model", "thread-123");
+    await endpointFetch(endpoint.baseUrl, {
+      method: "POST",
+      headers: { "content-type": "application/json" },
+      body: JSON.stringify({ messages: [] }),
+    });
+
+    expect(capturedHeaders?.get("user-agent")).toBe("ReadAny/1.0");
+    expect(capturedHeaders?.get("x-opencode-session")).toBe("thread-123");
+  });
+
+  it("sends a fallback session id for non-conversation requests", async () => {
+    let capturedHeaders: Headers | undefined;
+    globalThis.fetch = vi.fn(async (input: RequestInfo | URL, init?: RequestInit) => {
+      capturedHeaders = input instanceof Request ? input.headers : new Headers(init?.headers);
+      return new Response("{}", {
+        status: 200,
+        headers: { "content-type": "application/json" },
+      });
+    }) as typeof fetch;
+
+    const endpoint = makeEndpoint({
+      provider: "opencode",
+      baseUrl: "https://opencode.ai/zen/go/v1/chat/completions",
+      useExactRequestUrl: false,
+    });
+    const endpointFetch = getEndpointFetch(endpoint, "some-model");
+    await endpointFetch(endpoint.baseUrl, {
+      method: "POST",
+      headers: { "content-type": "application/json" },
+      body: JSON.stringify({ messages: [] }),
+    });
+
+    expect(capturedHeaders?.get("x-opencode-session")).toMatch(/^readany-/);
+  });
+});
+
 describe("getEndpointFetch custom endpoint compatibility", () => {
   it("retries without token limit when a custom endpoint rejects max_completion_tokens over 100", async () => {
     const requestBodies: Record<string, unknown>[] = [];

packages/core/src/ai/agents/reading-agent.ts

@@ -592,6 +592,8 @@ export interface ReadingAgentOptions {
   }) => ToolDefinition[];
   /** Abort signal for immediate cancellation */
   signal?: AbortSignal;
+  /** Stable conversation/session id, forwarded to the provider (e.g. OpenCode Go routing). */
+  sessionId?: string;
   /** Maximum time a single tool may run before returning an error result. */
   toolTimeoutMs?: number;
 }
@@ -818,6 +820,7 @@ export async function* streamReadingAgent(
     memorySummary,
     getAvailableTools,
     signal,
+    sessionId,
     toolTimeoutMs = DEFAULT_TOOL_TIMEOUT_MS,
   } = options;
 
@@ -863,6 +866,7 @@ export async function* streamReadingAgent(
       maxTokens: aiConfig.maxTokens,
       streaming: true,
       deepThinking,
+      sessionId,
     });
 
     // Check abort after async operation

packages/core/src/ai/llm-provider.ts

@@ -33,6 +33,42 @@ function shouldSanitizeCustomHeaders(endpoint: AIEndpoint): boolean {
   return endpoint.provider === "custom";
 }
 
+/**
+ * OpenCode Go monitors traffic and asks clients to identify themselves with a
+ * dedicated user agent (not a generic SDK/HTTP library name) and to send a
+ * stable session id per conversation so it can optimize routing/prompt caching.
+ */
+const OPENCODE_PROVIDER_ID = "opencode";
+const READANY_CLIENT_USER_AGENT = "ReadAny/1.0";
+
+/**
+ * OpenCode Go expects the x-opencode-session header on every request, including
+ * non-conversation calls like connection tests and model listing. Fall back to a
+ * per-fetch-instance id when no conversation id was provided.
+ */
+function createOpencodeSessionId(): string {
+  return `readany-${Date.now().toString(36)}-${Math.random().toString(36).slice(2, 10)}`;
+}
+
+function applyOpencodeHeaders(
+  input: RequestInfo | URL,
+  init: RequestInit | undefined,
+  sessionId?: string,
+): { input: RequestInfo | URL; init?: RequestInit } {
+  const headers = mergeRequestHeaders(input, init) ?? new Headers();
+  headers.set("user-agent", READANY_CLIENT_USER_AGENT);
+  if (sessionId) headers.set("x-opencode-session", sessionId);
+
+  if (isRequestLike(input)) {
+    return {
+      input: new Request(input, { headers }),
+      init: init ? { ...init, headers } : undefined,
+    };
+  }
+
+  return { input, init: { ...(init ?? {}), headers } };
+}
+
 function mergeRequestHeaders(input: RequestInfo | URL, init?: RequestInit): Headers | undefined {
   const merged = new Headers();
   let hasHeaders = false;
@@ -375,9 +411,15 @@ async function patchGeminiThoughtSignatureRequest(
   };
 }
 
-export function getEndpointFetch(endpoint: AIEndpoint, model?: string): typeof globalThis.fetch {
+export function getEndpointFetch(
+  endpoint: AIEndpoint,
+  model?: string,
+  sessionId?: string,
+): typeof globalThis.fetch {
   const exactUrl = endpoint.useExactRequestUrl ? endpoint.baseUrl?.trim() : "";
   const baseFetch = (_streamingFetch ?? globalThis.fetch).bind(globalThis);
+  const opencodeSessionId =
+    endpoint.provider === OPENCODE_PROVIDER_ID ? (sessionId ?? createOpencodeSessionId()) : undefined;
 
   return (async (input: RequestInfo | URL, init?: RequestInit) => {
     const finalInput = isRequestLike(input)
@@ -403,6 +445,12 @@ export function getEndpointFetch(endpoint: AIEndpoint, model?: string): typeof g
       }
     }
 
+    if (endpoint.provider === OPENCODE_PROVIDER_ID) {
+      const withOpencodeHeaders = applyOpencodeHeaders(requestInput, requestInit, opencodeSessionId);
+      requestInput = withOpencodeHeaders.input;
+      requestInit = withOpencodeHeaders.init;
+    }
+
     if (
       requestMethod.toUpperCase() === "POST" &&
       shouldPatchGeminiThoughtSignatures(endpoint, model, requestUrl)
@@ -542,6 +590,8 @@ export interface LLMOptions {
   maxTokens?: number;
   streaming?: boolean;
   deepThinking?: boolean;
+  /** Stable conversation/session id (sent as x-opencode-session for OpenCode Go). */
+  sessionId?: string;
 }
 
 export function resolveActiveEndpoint(config: AIConfig): {
@@ -581,6 +631,7 @@ export async function createChatModel(
     maxTokens: options.maxTokens ?? config.maxTokens,
     streaming: options.streaming,
     deepThinking: options.deepThinking,
+    sessionId: options.sessionId,
   });
 }
 
@@ -601,7 +652,7 @@ export async function createChatModelFromEndpoint(
   const temperature = options.temperature ?? 0.7;
   const maxTokens = options.maxTokens ?? 8192;
   const streaming = options.streaming ?? true;
-  const endpointFetch = getEndpointFetch(endpoint, model);
+  const endpointFetch = getEndpointFetch(endpoint, model, options.sessionId);
 
   switch (endpoint.provider) {
     case "anthropic": {

packages/core/src/ai/streaming.ts

@@ -113,6 +113,7 @@ export class StreamingChat {
           memorySummary: options.thread.memorySummary,
           getAvailableTools: options.getAvailableTools,
           signal,
+          sessionId: options.thread.id,
         },
         userInput,
         history,

packages/core/src/types/chat.ts

@@ -145,6 +145,7 @@ export type AIProviderType =
   | "mistral"
   | "perplexity"
   | "aihubmix"
+  | "opencode"
   | "custom";
 
 export interface AIEndpoint {

packages/core/src/utils/api.ts

@@ -201,6 +201,14 @@ export const PROVIDER_CONFIGS: Record<string, ProviderConfig> = {
     placeholder: "https://aihubmix.com",
     keyPlaceholder: "sk-...",
   },
+  opencode: {
+    id: "opencode",
+    name: "OpenCode Go",
+    defaultBaseUrl: "https://opencode.ai/zen/go/v1",
+    needsV1Suffix: false,
+    placeholder: "https://opencode.ai/zen/go/v1",
+    keyPlaceholder: "sk-...",
+  },
   custom: {
     id: "custom",
     name: "Custom (OpenAI Compatible)",
@@ -566,6 +574,7 @@ export function detectProviderFromUrl(url: string): string {
   if (urlLower.includes("mistral.ai")) return "mistral";
   if (urlLower.includes("perplexity.ai")) return "perplexity";
   if (urlLower.includes("aihubmix.com")) return "aihubmix";
+  if (urlLower.includes("opencode.ai")) return "opencode";
 
   return "custom";
 }