Merge pull request #196 from bramses/settings-improvements

adds z.ai Provider and simplifies the settings UI

Author: DenizOkcu

package.json

@@ -21,23 +21,23 @@
   "keywords": [],
   "author": "",
   "dependencies": {
-    "@ai-sdk/anthropic": "^3.0.35",
-    "@ai-sdk/google": "^3.0.20",
-    "@ai-sdk/openai": "^3.0.25",
-    "@ai-sdk/openai-compatible": "^2.0.26",
+    "@ai-sdk/anthropic": "^3.0.38",
+    "@ai-sdk/google": "^3.0.22",
+    "@ai-sdk/openai": "^3.0.26",
+    "@ai-sdk/openai-compatible": "^2.0.28",
     "@openrouter/ai-sdk-provider": "^2.1.1",
-    "ai": "6.0.67",
+    "ai": "6.0.77",
     "zod": "4.3.6"
   },
   "devDependencies": {
-    "@codemirror/state": "^6.5.4",
-    "@codemirror/view": "^6.39.12",
-    "@types/node": "^25.2.0",
+    "@codemirror/state": "6.5.0",
+    "@codemirror/view": "6.38.6",
+    "@types/node": "^25.2.2",
     "@typescript-eslint/eslint-plugin": "8.54.0",
     "@typescript-eslint/parser": "8.54.0",
     "builtin-modules": "5.0.0",
-    "esbuild": "0.27.2",
-    "eslint": "^9.39.2",
+    "esbuild": "0.27.3",
+    "eslint": "^9.0.0",
     "obsidian": "latest",
     "prettier": "^3.8.1",
     "tslib": "2.8.1",

src/Commands/CommandUtilities.ts

@@ -9,6 +9,7 @@ import {
   AI_SERVICE_OLLAMA,
   AI_SERVICE_OPENAI,
   AI_SERVICE_OPENROUTER,
+  AI_SERVICE_ZAI,
   FETCH_MODELS_TIMEOUT_MS,
 } from "src/Constants";
 import { getApiUrlsFromFrontmatter } from "src/Utilities/FrontmatterHelpers";
@@ -19,6 +20,7 @@ import {
   DEFAULT_OLLAMA_CONFIG,
   DEFAULT_OPENAI_CONFIG,
   DEFAULT_OPENROUTER_CONFIG,
+  DEFAULT_ZAI_CONFIG,
 } from "src/Services/DefaultConfigs";
 
 /**
@@ -40,6 +42,7 @@ export function getDefaultApiUrls(settings: ChatGPT_MDSettings): { [key: string]
     [AI_SERVICE_LMSTUDIO]: settings.lmstudioUrl || DEFAULT_LMSTUDIO_CONFIG.url,
     [AI_SERVICE_ANTHROPIC]: settings.anthropicUrl || DEFAULT_ANTHROPIC_CONFIG.url,
     [AI_SERVICE_GEMINI]: settings.geminiUrl || DEFAULT_GEMINI_CONFIG.url,
+    [AI_SERVICE_ZAI]: settings.zaiUrl || DEFAULT_ZAI_CONFIG.url,
   };
 }
 
@@ -140,6 +143,18 @@ export async function fetchAvailableModels(
       );
     }
 
+    // Conditionally add Z.AI promise
+    const zaiApiKey = apiAuthService.getApiKey(settingsService.getSettings(), AI_SERVICE_ZAI);
+    if (isValidApiKey(zaiApiKey)) {
+      promises.push(
+        withTimeout(
+          aiService.fetchAvailableModels(urls[AI_SERVICE_ZAI], zaiApiKey, settingsService.getSettings(), "zai"),
+          FETCH_MODELS_TIMEOUT_MS,
+          []
+        )
+      );
+    }
+
     // Fetch all models in parallel and flatten the results
     const results = await Promise.all(promises);
     return results.flat();

src/Commands/ModelSelectHandler.ts

@@ -8,7 +8,9 @@ import {
   AI_SERVICE_OLLAMA,
   AI_SERVICE_OPENAI,
   AI_SERVICE_OPENROUTER,
+  AI_SERVICE_ZAI,
 } from "src/Constants";
+import { DEFAULT_ZAI_CONFIG } from "src/Services/DefaultConfigs";
 import { fetchAvailableModels, getAiApiUrls } from "./CommandUtilities";
 
 /**
@@ -61,6 +63,8 @@ export class ModelSelectHandler {
           [AI_SERVICE_ANTHROPIC]:
             frontmatter.anthropicUrl || settings.anthropicUrl || getAiApiUrls(frontmatter).anthropic,
           [AI_SERVICE_GEMINI]: frontmatter.geminiUrl || settings.geminiUrl || getAiApiUrls(frontmatter).gemini,
+          [AI_SERVICE_ZAI]:
+            frontmatter.zaiUrl || settings.zaiUrl || getAiApiUrls(frontmatter).zai || DEFAULT_ZAI_CONFIG.url,
         };
 
         const aiService = this.services.aiProviderService();
@@ -115,6 +119,7 @@ export class ModelSelectHandler {
         [AI_SERVICE_LMSTUDIO]: settings.lmstudioUrl || getAiApiUrls({}).lmstudio,
         [AI_SERVICE_ANTHROPIC]: settings.anthropicUrl || getAiApiUrls({}).anthropic,
         [AI_SERVICE_GEMINI]: settings.geminiUrl || getAiApiUrls({}).gemini,
+        [AI_SERVICE_ZAI]: settings.zaiUrl || getAiApiUrls({}).zai || DEFAULT_ZAI_CONFIG.url,
       };
 
       const aiService = this.services.aiProviderService();

src/Constants.ts

@@ -9,6 +9,7 @@ export const AI_SERVICE_OPENROUTER = "openrouter";
 export const AI_SERVICE_LMSTUDIO = "lmstudio";
 export const AI_SERVICE_ANTHROPIC = "anthropic";
 export const AI_SERVICE_GEMINI = "gemini";
+export const AI_SERVICE_ZAI = "zai";
 
 // API endpoints for each service
 export const API_ENDPOINTS = {

src/Models/Config.ts

@@ -11,6 +11,7 @@ import {
   DEFAULT_OLLAMA_CONFIG,
   DEFAULT_OPENAI_CONFIG,
   DEFAULT_OPENROUTER_CONFIG,
+  DEFAULT_ZAI_CONFIG,
 } from "src/Services/DefaultConfigs";
 import { getDefaultToolWhitelist } from "src/Services/ToolSupportDetector";
 
@@ -49,6 +50,8 @@ export interface ApiKeySettings {
   anthropicApiKey: string;
   /** API Key for Gemini - used for Google Gemini models */
   geminiApiKey: string;
+  /** API Key for Z.AI - used for GLM models (both Standard API and Coding Plan) */
+  zaiApiKey: string;
 }
 
 /**
@@ -154,6 +157,18 @@ export interface LmStudioFrontmatterSettings {
   lmstudioDefaultFrequencyPenalty: number;
 }
 
+/**
+ * Provider-specific frontmatter settings for Z.AI
+ */
+export interface ZaiFrontmatterSettings {
+  /** Default model for Z.AI chats */
+  zaiDefaultModel: string;
+  /** Default temperature for Z.AI chats */
+  zaiDefaultTemperature: number;
+  /** Default max tokens for Z.AI chats */
+  zaiDefaultMaxTokens: number;
+}
+
 /**
  * Chat template settings
  */
@@ -178,6 +193,8 @@ export interface ServiceUrlSettings {
   anthropicUrl: string;
   /** URL for Gemini API */
   geminiUrl: string;
+  /** URL for Z.AI API (Standard: /api/paas/v4, Coding Plan: /api/anthropic) */
+  zaiUrl: string;
 }
 
 /**
@@ -211,7 +228,8 @@ export interface ChatGPT_MDSettings
     GeminiFrontmatterSettings,
     OpenRouterFrontmatterSettings,
     OllamaFrontmatterSettings,
-    LmStudioFrontmatterSettings {}
+    LmStudioFrontmatterSettings,
+    ZaiFrontmatterSettings {}
 
 /**
  * Default settings
@@ -222,6 +240,7 @@ export const DEFAULT_SETTINGS: ChatGPT_MDSettings = {
   openrouterApiKey: "",
   anthropicApiKey: "",
   geminiApiKey: "",
+  zaiApiKey: "",
 
   // Service URLs
   openaiUrl: DEFAULT_OPENAI_CONFIG.url,
@@ -230,6 +249,7 @@ export const DEFAULT_SETTINGS: ChatGPT_MDSettings = {
   lmstudioUrl: DEFAULT_LMSTUDIO_CONFIG.url,
   anthropicUrl: DEFAULT_ANTHROPIC_CONFIG.url,
   geminiUrl: DEFAULT_GEMINI_CONFIG.url,
+  zaiUrl: DEFAULT_ZAI_CONFIG.url,
 
   // Folders
   chatFolder: "ChatGPT_MD/chats",
@@ -294,4 +314,9 @@ export const DEFAULT_SETTINGS: ChatGPT_MDSettings = {
   lmstudioDefaultTopP: DEFAULT_LMSTUDIO_CONFIG.top_p,
   lmstudioDefaultPresencePenalty: DEFAULT_LMSTUDIO_CONFIG.presence_penalty,
   lmstudioDefaultFrequencyPenalty: DEFAULT_LMSTUDIO_CONFIG.frequency_penalty,
+
+  // Z.AI Defaults
+  zaiDefaultModel: DEFAULT_ZAI_CONFIG.model,
+  zaiDefaultTemperature: DEFAULT_ZAI_CONFIG.temperature,
+  zaiDefaultMaxTokens: DEFAULT_ZAI_CONFIG.max_tokens,
 };

src/Services/Adapters/AnthropicAdapter.ts

@@ -43,7 +43,8 @@ export class AnthropicAdapter extends BaseProviderAdapter {
     }
 
     try {
-      const modelsUrl = `${url.replace(/\/$/, "")}/v1/models`;
+      const apiPath = this.getApiPathSuffix(url);
+      const modelsUrl = `${url.replace(/\/$/, "")}${apiPath}/models`;
       const headers = this.getAuthHeaders(apiKey!); // Non-null assertion: validated above
 
       // Anthropic uses requestUrl directly (not through apiService)

src/Services/Adapters/BaseProviderAdapter.ts

@@ -101,4 +101,13 @@ export abstract class BaseProviderAdapter implements ProviderAdapter {
     }
     return true;
   }
+
+  /**
+   * Default API path suffix for chat completions
+   * Most OpenAI-compatible providers use "/v1"
+   * @param url - Optional URL parameter (ignored by most providers)
+   */
+  getApiPathSuffix(_url?: string): string {
+    return "/v1";
+  }
 }

src/Services/Adapters/CLAUDE.md

@@ -13,7 +13,7 @@ Each adapter encapsulates provider-specific logic, allowing `AiProviderService`
 ### ProviderType
 
 ```typescript
-type ProviderType = "openai" | "anthropic" | "ollama" | "openrouter" | "gemini" | "lmstudio";
+type ProviderType = "openai" | "anthropic" | "ollama" | "openrouter" | "gemini" | "lmstudio" | "zai";
 ```
 
 ### AiProviderConfig
@@ -51,19 +51,22 @@ Contract each adapter must implement:
 - `supportsToolCalling()` - Tool calling support
 - `requiresApiKey()` - Local providers: false
 - `extractModelName(modelId)` - Strip provider prefix
+- `getApiPathSuffix()` - API path suffix for chat completions (e.g., "/v1", "/api/v1", "")
 
 ## BaseProviderAdapter.ts
 
 **Abstract base class with common functionality**
 
 Default implementations:
+
 - `getSystemMessageRole()` → "system"
 - `supportsSystemField()` → false
 - `supportsToolCalling()` → true
 - `requiresApiKey()` → true
 - `extractModelName()` - Remove provider prefix
 - `validateApiKey()` - Check key presence
 - `handleFetchError()` - Consistent error logging
+- `getApiPathSuffix()` → "/v1" (default for most OpenAI-compatible providers)
 
 ## Adapter Implementations
 
@@ -99,5 +102,20 @@ Default implementations:
 ### OpenRouterAdapter.ts
 
 - Default URL: `https://openrouter.ai`
+- `getApiPathSuffix()` → "/api/v1" (OpenRouter's unique structure)
 - Auth: `Authorization: Bearer {key}`
 - Prefixes models with `openrouter@`
+- Final endpoints: `/api/v1/chat/completions`, `/api/v1/models`
+
+### ZaiAdapter.ts
+
+- Default URL: `https://api.z.ai`
+- Uses `createOpenAICompatible` from AI SDK
+- `getApiPathSuffix(url)` → `/api/paas/v4` (Standard mode) or `/api/anthropic/v1` (Coding Plan mode)
+- Supports two API modes based on URL path:
+  - Standard API: Uses `createOpenAICompatible` with `/api/paas/v4` path
+  - Coding Plan (Anthropic-compatible): Uses `createOpenAICompatible` with `/api/anthropic/v1` path
+- Auth: `Authorization: Bearer {key}`
+- No models endpoint - returns known models directly:
+  - GLM-4.5, GLM-4.6, GLM-4.6V, GLM-4.6V-Flash, GLM-4.6V-FlashX, GLM-4.7, GLM-4.7-Flash
+- Prefixes models with `zai@`

src/Services/Adapters/OpenRouterAdapter.ts

@@ -35,6 +35,14 @@ export class OpenRouterAdapter extends BaseProviderAdapter {
     };
   }
 
+  /**
+   * OpenRouter requires /api/v1 prefix for all endpoints
+   * @param url - Optional URL parameter (ignored by OpenRouter)
+   */
+  override getApiPathSuffix(_url?: string): string {
+    return "/api/v1";
+  }
+
   async fetchModels(
     url: string,
     apiKey: string | undefined,
@@ -47,7 +55,8 @@ export class OpenRouterAdapter extends BaseProviderAdapter {
 
     try {
       const headers = this.getAuthHeaders(apiKey!); // Non-null assertion: validated above
-      const models = await makeGetRequest(`${url}/api/v1/models`, headers, this.type);
+      const apiPath = this.getApiPathSuffix(url);
+      const models = await makeGetRequest(`${url}${apiPath}/models`, headers, this.type);
 
       return models.data
         .sort((a: OpenRouterModel, b: OpenRouterModel) => {

src/Services/Adapters/ProviderAdapter.ts

@@ -3,7 +3,7 @@ import { ChatGPT_MDSettings } from "src/Models/Config";
 /**
  * Union type for all supported AI providers
  */
-export type ProviderType = "openai" | "anthropic" | "ollama" | "openrouter" | "gemini" | "lmstudio";
+export type ProviderType = "openai" | "anthropic" | "ollama" | "openrouter" | "gemini" | "lmstudio" | "zai";
 
 /**
  * Unified configuration interface for all AI providers
@@ -107,4 +107,13 @@ export interface ProviderAdapter {
    * @param modelId - Model ID with or without provider prefix
    */
   extractModelName(modelId: string): string;
+
+  /**
+   * Get the API path suffix for chat completions endpoint
+   * Most OpenAI-compatible providers use "/v1"
+   * OpenRouter uses "/api/v1" to handle its unique structure
+   * Z.AI returns "/api/paas/v4" or "/api/anthropic/v1" depending on the URL
+   * @param url - Optional URL to determine the correct suffix (used by Z.AI for mode detection)
+   */
+  getApiPathSuffix(url?: string): string;
 }