feat(api): add MiniMax provider option to the ontology builder

README.md

@@ -207,12 +207,28 @@ GitHub Pages build so asset paths resolve correctly.
 
 | Variable | Default | Description |
 |----------|---------|-------------|
-| `VITE_ENABLE_AI_BUILDER` | `false` | Enable the Azure OpenAI ontology builder |
+| `VITE_ENABLE_AI_BUILDER` | `false` | Enable the AI ontology builder (backed by the `api/` function) |
 | `VITE_ENABLE_LEGACY_FORMATS` | `false` | Enable JSON/YAML/CSV import/export formats |
 | `VITE_BASE_PATH` | `/` | Base path for the app (set automatically for GitHub Pages) |
 | `VITE_GITHUB_CLIENT_ID` | *(empty)* | GitHub OAuth App client ID for one-click catalogue PRs ([setup guide](docs/github-oauth-setup.md)) |
 | `VITE_GITHUB_OAUTH_BASE` | *(empty)* | External OAuth proxy URL for GitHub Pages deployments (e.g. Cloudflare Worker URL) |
 
+#### AI builder backend (`api/`)
+
+The `generate-ontology` function selects its language model provider from
+`AI_PROVIDER` (default `azure`). Configure these in `api/local.settings.json`
+(local) or the Function App application settings (deployed).
+
+| Variable | Default | Description |
+|----------|---------|-------------|
+| `AI_PROVIDER` | `azure` | Provider for the ontology builder: `azure` or `minimax` |
+| `AZURE_OPENAI_ENDPOINT` | *(empty)* | Azure OpenAI resource endpoint (required when `AI_PROVIDER=azure`) |
+| `AZURE_OPENAI_API_KEY` | *(empty)* | Azure OpenAI API key (required when `AI_PROVIDER=azure`) |
+| `AZURE_OPENAI_DEPLOYMENT` | `gpt-4o-mini` | Azure OpenAI deployment name |
+| `MINIMAX_BASE_URL` | `https://api.minimax.io/v1` | MiniMax OpenAI-compatible base URL (use `https://api.minimaxi.com/v1` for mainland China) |
+| `MINIMAX_API_KEY` | *(empty)* | MiniMax API key (required when `AI_PROVIDER=minimax`) |
+| `MINIMAX_MODEL` | `MiniMax-M3` | MiniMax model id |
+
 ## Project Structure
 
 ```

api/generate-ontology/index.ts

@@ -8,6 +8,15 @@ interface OpenAIResponse {
   }>;
 }
 
+// Resolved chat-completions call for the selected provider.
+interface ProviderRequest {
+  url: string;
+  headers: Record<string, string>;
+  // Model identifier sent in the request body. Azure OpenAI selects the model
+  // through the deployment segment of the URL and omits this field.
+  model?: string;
+}
+
 const SYSTEM_PROMPT = `You are an expert ontology extraction system. Given a business scenario description, extract entities, relationships, and properties to create a complete ontology.
 
 Output ONLY valid JSON matching this exact schema:
@@ -53,6 +62,62 @@ Rules:
 7. Assign unique hex colors to each entity (use Microsoft palette: #0078D4, #107C10, #5C2D91, #FFB900, #D83B01, #00A9E0, #8764B8, #00B294)
 8. Output ONLY the JSON, no explanations`;
 
+// Build the chat-completions request for the configured provider. Azure OpenAI
+// remains the default so existing deployments keep working; set AI_PROVIDER to
+// switch to another OpenAI-compatible backend. Returns an error string when the
+// selected provider is missing required configuration.
+function resolveProvider(): { request?: ProviderRequest; error?: string } {
+  const provider = (process.env.AI_PROVIDER || "azure").trim().toLowerCase();
+
+  if (provider === "minimax") {
+    // MiniMax exposes an OpenAI-compatible chat-completions API. Use the global
+    // endpoint by default; set MINIMAX_BASE_URL to https://api.minimaxi.com/v1
+    // for the mainland China endpoint.
+    const baseUrl = (process.env.MINIMAX_BASE_URL || "https://api.minimax.io/v1").replace(/\/+$/, "");
+    const apiKey = process.env.MINIMAX_API_KEY;
+    const model = process.env.MINIMAX_MODEL || "MiniMax-M3";
+
+    if (!apiKey) {
+      return {
+        error:
+          "MiniMax not configured. Set MINIMAX_API_KEY (and optionally MINIMAX_BASE_URL and MINIMAX_MODEL).",
+      };
+    }
+
+    return {
+      request: {
+        url: `${baseUrl}/chat/completions`,
+        headers: {
+          "Content-Type": "application/json",
+          Authorization: `Bearer ${apiKey}`,
+        },
+        model,
+      },
+    };
+  }
+
+  // Default provider: Azure OpenAI.
+  const endpoint = process.env.AZURE_OPENAI_ENDPOINT;
+  const apiKey = process.env.AZURE_OPENAI_API_KEY;
+  const deployment = process.env.AZURE_OPENAI_DEPLOYMENT || "gpt-4o-mini";
+
+  if (!endpoint || !apiKey) {
+    return {
+      error: "Azure OpenAI not configured. Set AZURE_OPENAI_ENDPOINT and AZURE_OPENAI_API_KEY.",
+    };
+  }
+
+  return {
+    request: {
+      url: `${endpoint}openai/deployments/${deployment}/chat/completions?api-version=2024-02-15-preview`,
+      headers: {
+        "Content-Type": "application/json",
+        "api-key": apiKey,
+      },
+    },
+  };
+}
+
 const generateOntology: AzureFunction = async function (
   context: Context,
   req: HttpRequest
@@ -67,45 +132,45 @@ const generateOntology: AzureFunction = async function (
     return;
   }
 
-  const endpoint = process.env.AZURE_OPENAI_ENDPOINT;
-  const apiKey = process.env.AZURE_OPENAI_API_KEY;
-  const deployment = process.env.AZURE_OPENAI_DEPLOYMENT || "gpt-4o-mini";
+  const { request: providerRequest, error: providerError } = resolveProvider();
 
-  if (!endpoint || !apiKey) {
+  if (!providerRequest) {
     context.res = {
       status: 500,
-      body: { error: "Azure OpenAI not configured. Set AZURE_OPENAI_ENDPOINT and AZURE_OPENAI_API_KEY." },
+      body: { error: providerError },
     };
     return;
   }
 
   try {
-    const response = await fetch(
-      `${endpoint}openai/deployments/${deployment}/chat/completions?api-version=2024-02-15-preview`,
-      {
-        method: "POST",
-        headers: {
-          "Content-Type": "application/json",
-          "api-key": apiKey,
-        },
-        body: JSON.stringify({
-          messages: [
-            { role: "system", content: SYSTEM_PROMPT },
-            { role: "user", content: description },
-          ],
-          temperature: 0.3,
-          max_tokens: 4000,
-          response_format: { type: "json_object" },
-        }),
-      }
-    );
+    const requestBody: Record<string, unknown> = {
+      messages: [
+        { role: "system", content: SYSTEM_PROMPT },
+        { role: "user", content: description },
+      ],
+      temperature: 0.3,
+      max_tokens: 4000,
+      response_format: { type: "json_object" },
+    };
+
+    // OpenAI-compatible providers select the model in the request body; Azure
+    // OpenAI encodes it in the deployment URL and leaves this undefined.
+    if (providerRequest.model) {
+      requestBody.model = providerRequest.model;
+    }
+
+    const response = await fetch(providerRequest.url, {
+      method: "POST",
+      headers: providerRequest.headers,
+      body: JSON.stringify(requestBody),
+    });
 
     if (!response.ok) {
       const errorText = await response.text();
-      context.log.error("Azure OpenAI error:", errorText);
+      context.log.error("Ontology provider error:", errorText);
       context.res = {
         status: 502,
-        body: { error: "Failed to generate ontology from Azure OpenAI" },
+        body: { error: "Failed to generate ontology from the language model provider" },
       };
       return;
     }
@@ -116,7 +181,7 @@ const generateOntology: AzureFunction = async function (
     if (!content) {
       context.res = {
         status: 500,
-        body: { error: "No content in Azure OpenAI response" },
+        body: { error: "No content in the language model response" },
       };
       return;
     }

api/local.settings.json.template

@@ -3,8 +3,12 @@
   "Values": {
     "FUNCTIONS_WORKER_RUNTIME": "node",
     "AzureWebJobsStorage": "",
+    "AI_PROVIDER": "azure",
     "AZURE_OPENAI_ENDPOINT": "https://YOUR-RESOURCE.openai.azure.com/",
     "AZURE_OPENAI_API_KEY": "YOUR-API-KEY",
-    "AZURE_OPENAI_DEPLOYMENT": "gpt-4o-mini"
+    "AZURE_OPENAI_DEPLOYMENT": "gpt-4o-mini",
+    "MINIMAX_BASE_URL": "https://api.minimax.io/v1",
+    "MINIMAX_API_KEY": "YOUR-API-KEY",
+    "MINIMAX_MODEL": "MiniMax-M3"
   }
 }