feat(ai): add MiniMax provider for global and China Anthropic Messages endpoints

README.md

@@ -344,6 +344,7 @@ Manual `memoryProvider` modes:
 - `openai-chat`: OpenAI Chat Completions compatible API with tool/function calling. This can work with compatible proxies such as LiteLLM only when the selected upstream model and proxy preserve tool calls.
 - `openai-responses`: OpenAI Responses API with function-call output.
 - `anthropic`: Anthropic Messages API with tool use.
+- `minimax`: MiniMax Anthropic Messages-compatible endpoint. Set `memoryApiUrl` to the global endpoint (`https://api.minimax.io`) or the China endpoint (`https://api.minimaxi.com`); the `/anthropic/v1/messages` path and `x-api-key` header are applied automatically. MiniMax text models such as `MiniMax-M3` support the adaptive thinking modes used by this plugin via `memoryExtraParams`.
 
 Troubleshooting:
 

src/config.ts

@@ -43,7 +43,7 @@ interface OpenCodeMemConfig {
   autoCaptureIterationTimeout?: number;
   autoCaptureMaxRetries?: number;
   autoCaptureLanguage?: string;
-  memoryProvider?: "openai-chat" | "openai-responses" | "anthropic";
+  memoryProvider?: "openai-chat" | "openai-responses" | "anthropic" | "minimax";
   memoryModel?: string;
   memoryApiUrl?: string;
   memoryApiKey?: string;
@@ -124,7 +124,7 @@ const DEFAULTS: Required<
   memoryModel?: string;
   memoryApiUrl?: string;
   memoryApiKey?: string;
-  memoryProvider?: "openai-chat" | "openai-responses" | "anthropic";
+  memoryProvider?: "openai-chat" | "openai-responses" | "anthropic" | "minimax";
   memoryTemperature?: number | false;
   memoryExtraParams?: Record<string, unknown>;
   opencodeProvider?: string;
@@ -343,7 +343,7 @@ const CONFIG_TEMPLATE = `{
 
   "autoCaptureEnabled": true,
 
-  // Provider type: "openai-chat" | "openai-responses" | "anthropic"
+  // Provider type: "openai-chat" | "openai-responses" | "anthropic" | "minimax"
   // Note: "openai-chat" is a generic OpenAI API-compatible mode.
   // Any service that follows the OpenAI Chat Completions API can use it via custom "memoryApiUrl".
   "memoryProvider": "openai-chat",
@@ -386,14 +386,23 @@ const CONFIG_TEMPLATE = `{
   //   "memoryModel": "claude-3-5-haiku-20241022"
   //   "memoryApiUrl": "https://api.anthropic.com/v1"
   //   "memoryApiKey": "sk-ant-..."
-
+
+  // MiniMax (Anthropic Messages-compatible endpoint, with session support):
+  //   "memoryProvider": "minimax"
+  //   "memoryModel": "MiniMax-M3"
+  //   "memoryApiUrl": "https://api.minimax.io"        // global endpoint
+  //   "memoryApiKey": "<MiniMax API key>"
+  //   // China endpoint: "memoryApiUrl": "https://api.minimaxi.com"
+  //   // Optional adaptive thinking for MiniMax-M3:
+  //   "memoryExtraParams": { "thinking": { "type": "adaptive" } }
+
   // Groq (OpenAI-compatible, use openai-chat provider):
   //   "memoryProvider": "openai-chat"
   //   "memoryModel": "llama-3.3-70b-versatile"
   //   "memoryApiUrl": "https://api.groq.com/openai/v1"
   //   "memoryApiKey": "gsk_..."
 
-  // Maximum iterations for multi-turn AI analysis (for openai-responses and anthropic)
+  // Maximum iterations for multi-turn AI analysis (for openai-responses, anthropic, and minimax)
   "autoCaptureMaxIterations": 5,
 
   // Timeout per iteration in milliseconds (30 seconds default)
@@ -598,7 +607,7 @@ function buildConfig(fileConfig: OpenCodeMemConfig) {
     autoCaptureMaxRetries: fileConfig.autoCaptureMaxRetries ?? DEFAULTS.autoCaptureMaxRetries,
     autoCaptureLanguage: fileConfig.autoCaptureLanguage,
     memoryProvider: (fileConfig.memoryProvider ?? "openai-chat") as
-      "openai-chat" | "openai-responses" | "anthropic",
+      "openai-chat" | "openai-responses" | "anthropic" | "minimax",
     memoryModel: fileConfig.memoryModel,
     memoryApiUrl: fileConfig.memoryApiUrl,
     memoryApiKey,

src/services/ai/ai-provider-factory.ts

@@ -2,6 +2,7 @@ import { BaseAIProvider, type ProviderConfig } from "./providers/base-provider.j
 import { OpenAIChatCompletionProvider } from "./providers/openai-chat-completion.js";
 import { OpenAIResponsesProvider } from "./providers/openai-responses.js";
 import { AnthropicMessagesProvider } from "./providers/anthropic-messages.js";
+import { MiniMaxProvider } from "./providers/minimax.js";
 import { GoogleGeminiProvider } from "./providers/google-gemini.js";
 import { aiSessionManager } from "./session/ai-session-manager.js";
 import type { AIProviderType } from "./session/session-types.js";
@@ -18,6 +19,9 @@ export class AIProviderFactory {
       case "anthropic":
         return new AnthropicMessagesProvider(config, aiSessionManager);
 
+      case "minimax":
+        return new MiniMaxProvider(config, aiSessionManager);
+
       case "google-gemini":
         return new GoogleGeminiProvider(config, aiSessionManager);
 
@@ -27,7 +31,7 @@ export class AIProviderFactory {
   }
 
   static getSupportedProviders(): AIProviderType[] {
-    return ["openai-chat", "openai-responses", "anthropic", "google-gemini"];
+    return ["openai-chat", "openai-responses", "anthropic", "minimax", "google-gemini"];
   }
 
   static async cleanupExpiredSessions(): Promise<number> {

src/services/ai/providers/anthropic-messages.ts

@@ -1,6 +1,7 @@
-import { BaseAIProvider, type ToolCallResult } from "./base-provider.js";
+import { applySafeExtraParams, BaseAIProvider, type ToolCallResult } from "./base-provider.js";
 import { AISessionManager } from "../session/ai-session-manager.js";
 import { ToolSchemaConverter, type ChatCompletionTool } from "../tools/tool-schema.js";
+import type { AIProviderType } from "../session/session-types.js";
 import { log } from "../../logger.js";
 import { UserProfileValidator } from "../validators/user-profile-validator.js";
 
@@ -28,6 +29,14 @@ interface AnthropicResponse {
   };
 }
 
+/**
+ * Base implementation for providers that speak the Anthropic Messages API
+ * (Anthropic itself plus Anthropic-compatible gateways such as MiniMax).
+ *
+ * Subclasses override the session provider tag and the resolved endpoint URL so
+ * the session store and request routing reflect the upstream provider while the
+ * message/tool-call handling stays shared.
+ */
 export class AnthropicMessagesProvider extends BaseAIProvider {
   private aiSessionManager: AISessionManager;
 
@@ -44,17 +53,47 @@ export class AnthropicMessagesProvider extends BaseAIProvider {
     return true;
   }
 
+  /**
+   * Provider tag stored on AI sessions so a subclass (e.g. MiniMax) records its
+   * own tag instead of the literal "anthropic" value.
+   */
+  protected sessionProviderTag(): AIProviderType {
+    return "anthropic";
+  }
+
+  /**
+   * Resolve the Messages endpoint URL. The default appends `/messages` to the
+   * configured base URL, matching Anthropic's `https://api.anthropic.com/v1`
+   * base. Subclasses with a different path layout override this.
+   */
+  protected resolveEndpoint(): string {
+    return `${this.config.apiUrl}/messages`;
+  }
+
+  protected apiErrorLogLabel(): string {
+    return "Anthropic Messages API error";
+  }
+
+  protected toolValidationErrorLogLabel(): string {
+    return "Anthropic tool response validation failed";
+  }
+
+  protected timeoutLabel(): string {
+    return "Anthropic API request timeout";
+  }
+
   async executeToolCall(
     systemPrompt: string,
     userPrompt: string,
     toolSchema: ChatCompletionTool,
     sessionId: string
   ): Promise<ToolCallResult> {
-    let session = await this.aiSessionManager.getSession(sessionId, "anthropic");
+    const providerTag = this.sessionProviderTag();
+    let session = await this.aiSessionManager.getSession(sessionId, providerTag);
 
     if (!session) {
       session = await this.aiSessionManager.createSession({
-        provider: "anthropic",
+        provider: providerTag,
         sessionId,
         metadata: { systemPrompt },
       });
@@ -97,14 +136,18 @@ export class AnthropicMessagesProvider extends BaseAIProvider {
       try {
         const tool = ToolSchemaConverter.toAnthropic(toolSchema);
 
-        const requestBody = {
+        const requestBody: Record<string, unknown> = {
           model: this.config.model,
           max_tokens: this.config.maxTokens ?? 4096,
           system: systemPrompt,
           messages,
           tools: [tool],
         };
 
+        if (this.config.extraParams) {
+          applySafeExtraParams(requestBody, this.config.extraParams);
+        }
+
         const headers: Record<string, string> = {
           "Content-Type": "application/json",
           "anthropic-version": "2023-06-01",
@@ -114,7 +157,7 @@ export class AnthropicMessagesProvider extends BaseAIProvider {
           headers["x-api-key"] = this.config.apiKey;
         }
 
-        const response = await fetch(`${this.config.apiUrl}/messages`, {
+        const response = await fetch(this.resolveEndpoint(), {
           method: "POST",
           headers,
           body: JSON.stringify(requestBody),
@@ -125,7 +168,7 @@ export class AnthropicMessagesProvider extends BaseAIProvider {
 
         if (!response.ok) {
           const errorText = await response.text().catch(() => response.statusText);
-          log("Anthropic Messages API error", {
+          log(this.apiErrorLogLabel(), {
             provider: this.getProviderName(),
             model: this.config.model,
             status: response.status,
@@ -170,7 +213,7 @@ export class AnthropicMessagesProvider extends BaseAIProvider {
             };
           } catch (validationError) {
             const errorStack = validationError instanceof Error ? validationError.stack : undefined;
-            log("Anthropic tool response validation failed", {
+            log(this.toolValidationErrorLogLabel(), {
               error: String(validationError),
               stack: errorStack,
               errorType:
@@ -210,7 +253,7 @@ export class AnthropicMessagesProvider extends BaseAIProvider {
         if (error instanceof Error && error.name === "AbortError") {
           return {
             success: false,
-            error: `API request timeout (${this.config.iterationTimeout}ms)`,
+            error: `${this.timeoutLabel()} (${this.config.iterationTimeout}ms)`,
             iterations,
           };
         }

src/services/ai/providers/minimax.ts

@@ -0,0 +1,63 @@
+import { type ProviderConfig } from "./base-provider.js";
+import { AISessionManager } from "../session/ai-session-manager.js";
+import { AnthropicMessagesProvider } from "./anthropic-messages.js";
+import type { AIProviderType } from "../session/session-types.js";
+
+/**
+ * MiniMax provider.
+ *
+ * MiniMax exposes an Anthropic Messages-compatible endpoint for its text
+ * models. The global endpoint (https://api.minimax.io) and the China endpoint
+ * (https://api.minimaxi.com) both expose the same `/anthropic/v1/messages`
+ * path and authenticate with the `x-api-key` header. This provider reuses the
+ * Anthropic Messages request/response handling and only overrides the resolved
+ * request endpoint URL and the session provider tag, so MiniMax is
+ * distinguishable from Anthropic in the session store and diagnostics.
+ *
+ * Users configure the base URL as `memoryApiUrl`:
+ *   - global endpoint: "https://api.minimax.io"
+ *   - China endpoint:  "https://api.minimaxi.com"
+ */
+export class MiniMaxProvider extends AnthropicMessagesProvider {
+  constructor(config: ProviderConfig, aiSessionManager: AISessionManager) {
+    super(config, aiSessionManager);
+  }
+
+  override getProviderName(): string {
+    return "minimax";
+  }
+
+  /**
+   * Resolve the Anthropic Messages endpoint URL for MiniMax.
+   *
+   * MiniMax's Messages endpoint lives at `<base>/anthropic/v1/messages`. The
+   * base URL is normalized so users can supply the host with or without a
+   * trailing `/anthropic`, `/anthropic/v1`, or `/anthropic/v1/messages` suffix.
+   */
+  override resolveEndpoint(): string {
+    let base = (this.config.apiUrl || "").trim().replace(/\/+$/, "");
+    if (!base) {
+      throw new Error("MiniMax provider requires a configured memoryApiUrl");
+    }
+    base = base.replace(/\/anthropic\/v1\/messages\/?$/, "");
+    base = base.replace(/\/anthropic\/v1\/?$/, "");
+    base = base.replace(/\/anthropic\/?$/, "");
+    return `${base}/anthropic/v1/messages`;
+  }
+
+  override sessionProviderTag(): AIProviderType {
+    return "minimax";
+  }
+
+  protected override apiErrorLogLabel(): string {
+    return "MiniMax Messages API error";
+  }
+
+  protected override toolValidationErrorLogLabel(): string {
+    return "MiniMax tool response validation failed";
+  }
+
+  protected override timeoutLabel(): string {
+    return "MiniMax API request timeout";
+  }
+}

src/services/ai/session/session-types.ts

@@ -1,4 +1,5 @@
-export type AIProviderType = "openai-chat" | "openai-responses" | "anthropic" | "google-gemini";
+export type AIProviderType =
+  "openai-chat" | "openai-responses" | "anthropic" | "minimax" | "google-gemini";
 
 export interface AIMessage {
   id?: number;

src/types/index.ts

@@ -17,4 +17,5 @@ export interface MemoryMetadata {
   [key: string]: unknown;
 }
 
-export type AIProviderType = "openai-chat" | "openai-responses" | "anthropic";
+export type AIProviderType =
+  "openai-chat" | "openai-responses" | "anthropic" | "minimax" | "google-gemini";