Merge b94b8aea56 into 09be5d45d5

2026-01-30 16:00:11 +00:00 · 2026-01-30 16:00:11 +00:00 · bc2b99f6b8
commit bc2b99f6b8
parent 09be5d45d5 b94b8aea56
7 changed files with 638 additions and 0 deletions
--- a/.env.example
+++ b/.env.example
@ -3,3 +3,15 @@ TWILIO_ACCOUNT_SID=ACxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
 TWILIO_AUTH_TOKEN=your_auth_token_here
 # Must be a WhatsApp-enabled Twilio number, prefixed with whatsapp:
 TWILIO_WHATSAPP_FROM=whatsapp:+17343367101
+
+# Azure OpenAI Configuration
+# Your Azure OpenAI API key
+AZURE_OPENAI_API_KEY=your_azure_openai_api_key_here
+# Your Azure OpenAI resource name (e.g., "my-openai-resource")
+AZURE_OPENAI_RESOURCE_NAME=your_resource_name_here
+# Your Azure OpenAI deployment name (e.g., "gpt-5", "gpt-5-codex")
+AZURE_OPENAI_DEPLOYMENT_NAME=your_deployment_name_here
+# API version (optional, defaults to 2024-08-01-preview)
+AZURE_OPENAI_API_VERSION=2024-08-01-preview
+# Or provide the full endpoint URL directly (alternative to resource name)
+AZURE_OPENAI_ENDPOINT=https://your-resource.openai.azure.com
--- a/docker-compose.azure.yml
+++ b/docker-compose.azure.yml
@ -0,0 +1,100 @@
+# Docker Compose for Azure OpenAI deployment
+# Usage:
+#   1. Copy .env.example to .env and fill in your Azure OpenAI credentials
+#   2. Build the image: docker compose -f docker-compose.azure.yml build
+#   3. Run the gateway: docker compose -f docker-compose.azure.yml up -d moltbot-azure-gateway
+#   4. Or run the CLI: docker compose -f docker-compose.azure.yml run --rm moltbot-azure-cli
+
+services:
+  moltbot-azure-gateway:
+    build:
+      context: .
+      dockerfile: Dockerfile
+    image: moltbot-azure:local
+    container_name: moltbot-azure-gateway
+    environment:
+      HOME: /home/node
+      TERM: xterm-256color
+      # Gateway authentication token (generate a secure token for production)
+      CLAWDBOT_GATEWAY_TOKEN: ${CLAWDBOT_GATEWAY_TOKEN:-your-secure-gateway-token}
+      # Azure OpenAI Configuration (required)
+      AZURE_OPENAI_API_KEY: ${AZURE_OPENAI_API_KEY:?Azure OpenAI API key is required}
+      AZURE_OPENAI_RESOURCE_NAME: ${AZURE_OPENAI_RESOURCE_NAME:?Azure OpenAI resource name is required}
+      AZURE_OPENAI_DEPLOYMENT_NAME: ${AZURE_OPENAI_DEPLOYMENT_NAME:?Azure OpenAI deployment name is required}
+      AZURE_OPENAI_API_VERSION: ${AZURE_OPENAI_API_VERSION:-2024-08-01-preview}
+      # Optional: provide full endpoint URL instead of resource name
+      AZURE_OPENAI_ENDPOINT: ${AZURE_OPENAI_ENDPOINT}
+    volumes:
+      # Persist configuration and session data
+      - ${CLAWDBOT_CONFIG_DIR:-./data/config}:/home/node/.moltbot
+      - ${CLAWDBOT_WORKSPACE_DIR:-./data/workspace}:/home/node/clawd
+    ports:
+      # Gateway HTTP/WebSocket port
+      - "${CLAWDBOT_GATEWAY_PORT:-18789}:18789"
+      # Bridge port for channel connections
+      - "${CLAWDBOT_BRIDGE_PORT:-18790}:18790"
+    init: true
+    restart: unless-stopped
+    command:
+      [
+        "node",
+        "dist/index.js",
+        "gateway",
+        "--bind",
+        "${CLAWDBOT_GATEWAY_BIND:-lan}",
+        "--port",
+        "${CLAWDBOT_GATEWAY_PORT:-18789}",
+        "--allow-unconfigured"
+      ]
+    healthcheck:
+      test: ["CMD", "curl", "-f", "http://localhost:18789/health"]
+      interval: 30s
+      timeout: 10s
+      retries: 3
+      start_period: 30s
+
+  moltbot-azure-cli:
+    build:
+      context: .
+      dockerfile: Dockerfile
+    image: moltbot-azure:local
+    container_name: moltbot-azure-cli
+    environment:
+      HOME: /home/node
+      TERM: xterm-256color
+      BROWSER: echo
+      # Azure OpenAI Configuration (required)
+      AZURE_OPENAI_API_KEY: ${AZURE_OPENAI_API_KEY:?Azure OpenAI API key is required}
+      AZURE_OPENAI_RESOURCE_NAME: ${AZURE_OPENAI_RESOURCE_NAME:?Azure OpenAI resource name is required}
+      AZURE_OPENAI_DEPLOYMENT_NAME: ${AZURE_OPENAI_DEPLOYMENT_NAME:?Azure OpenAI deployment name is required}
+      AZURE_OPENAI_API_VERSION: ${AZURE_OPENAI_API_VERSION:-2024-08-01-preview}
+      AZURE_OPENAI_ENDPOINT: ${AZURE_OPENAI_ENDPOINT}
+    volumes:
+      - ${CLAWDBOT_CONFIG_DIR:-./data/config}:/home/node/.moltbot
+      - ${CLAWDBOT_WORKSPACE_DIR:-./data/workspace}:/home/node/clawd
+    stdin_open: true
+    tty: true
+    init: true
+    entrypoint: ["node", "dist/index.js"]
+
+  # Minimal test service to verify Azure OpenAI connection
+  moltbot-azure-test:
+    build:
+      context: .
+      dockerfile: Dockerfile
+    image: moltbot-azure:local
+    environment:
+      HOME: /home/node
+      TERM: xterm-256color
+      AZURE_OPENAI_API_KEY: ${AZURE_OPENAI_API_KEY:?Azure OpenAI API key is required}
+      AZURE_OPENAI_RESOURCE_NAME: ${AZURE_OPENAI_RESOURCE_NAME:?Azure OpenAI resource name is required}
+      AZURE_OPENAI_DEPLOYMENT_NAME: ${AZURE_OPENAI_DEPLOYMENT_NAME:?Azure OpenAI deployment name is required}
+      AZURE_OPENAI_API_VERSION: ${AZURE_OPENAI_API_VERSION:-2024-08-01-preview}
+    command: ["node", "dist/index.js", "models", "list"]
+    profiles:
+      - test
+
+# Named volumes for persistent storage (optional, use bind mounts above for easier access)
+volumes:
+  moltbot-config:
+  moltbot-workspace:
--- a/docs/providers/azure-openai.md
+++ b/docs/providers/azure-openai.md
@ -0,0 +1,150 @@
+# Azure OpenAI Provider
+
+Configure Moltbot to use Azure OpenAI Service as a model provider.
+
+## Overview
+
+Azure OpenAI exposes OpenAI models deployed in your Azure subscription (for example `gpt-5` or `gpt-5-codex`) behind Azure endpoints and authentication.
+
+## Prerequisites
+
+1. An Azure subscription
+2. An Azure OpenAI resource created in Azure Portal
+3. At least one model deployment (e.g., `gpt-5` or `gpt-5-codex`)
+4. API key from the Azure OpenAI resource
+
+## Configuration
+
+### Environment Variables
+
+Set the following environment variables to configure Azure OpenAI:
+
+| Variable | Required | Description |
+|----------|----------|-------------|
+| `AZURE_OPENAI_API_KEY` | Yes | Your Azure OpenAI API key |
+| `AZURE_OPENAI_RESOURCE_NAME` | Yes* | Azure OpenAI resource name (e.g., `my-openai-resource`) |
+| `AZURE_OPENAI_DEPLOYMENT_NAME` | Yes | Model deployment name (e.g., `gpt-5`) |
+| `AZURE_OPENAI_API_VERSION` | No | API version (defaults to `2024-08-01-preview`) |
+| `AZURE_OPENAI_ENDPOINT` | No | Full endpoint URL (alternative to resource name) |
+
+\* Either `AZURE_OPENAI_RESOURCE_NAME` or `AZURE_OPENAI_ENDPOINT` is required.
+
+### Example .env Configuration
+
+```bash
+# Azure OpenAI Configuration
+AZURE_OPENAI_API_KEY=your_azure_openai_api_key_here
+AZURE_OPENAI_RESOURCE_NAME=my-openai-resource
+AZURE_OPENAI_DEPLOYMENT_NAME=gpt-5
+AZURE_OPENAI_API_VERSION=2024-08-01-preview
+
+# Alternative: provide full endpoint URL
+# AZURE_OPENAI_ENDPOINT=https://my-openai-resource.openai.azure.com
+```
+
+## Docker Deployment
+
+A dedicated Docker Compose file (`docker-compose.azure.yml`) is provided for Azure OpenAI deployments.
+
+### Quick Start
+
+1. Copy `.env.example` to `.env` and fill in your Azure OpenAI credentials:
+
+```bash
+cp .env.example .env
+# Edit .env with your Azure OpenAI configuration
+```
+
+2. Build the Docker image:
+
+```bash
+docker compose -f docker-compose.azure.yml build
+```
+
+3. Run the gateway:
+
+```bash
+docker compose -f docker-compose.azure.yml up -d moltbot-azure-gateway
+```
+
+4. Or run the CLI interactively:
+
+```bash
+docker compose -f docker-compose.azure.yml run --rm moltbot-azure-cli
+```
+
+### Services
+
+The `docker-compose.azure.yml` includes a gateway service, an interactive CLI service, and an optional test profile.
+
+### Testing Connection
+
+Verify your Azure OpenAI connection:
+
+```bash
+docker compose -f docker-compose.azure.yml --profile test run --rm moltbot-azure-test
+```
+
+## API Endpoint Format
+
+Azure OpenAI uses a different URL format than OpenAI:
+
+```
+https://{resourceName}.openai.azure.com/openai/deployments/{deploymentName}/chat/completions?api-version={apiVersion}
+```
+
+The provider automatically handles:
+- Building the correct endpoint URL from resource and deployment names
+- Adding the `api-version` query parameter via a global fetch wrapper
+- Using the `api-key` header instead of Bearer token authentication
+
+## Supported Models
+
+Moltbot treats Azure deployments as model IDs. Use your deployment name as `AZURE_OPENAI_DEPLOYMENT_NAME`.
+
+Recommended (newer) deployment names to use in docs/examples:
+
+- `gpt-5`
+- `gpt-5-mini`
+- `gpt-5-nano`
+- `gpt-5-codex`
+
+## Troubleshooting
+
+### Common Issues
+
+**401 Unauthorized**
+- Verify your `AZURE_OPENAI_API_KEY` is correct
+- Check that the API key has access to the specified deployment
+
+**404 Not Found**
+- Verify `AZURE_OPENAI_RESOURCE_NAME` matches your Azure resource
+- Verify `AZURE_OPENAI_DEPLOYMENT_NAME` matches an existing deployment
+- Check that the deployment is in a "Succeeded" state in Azure Portal
+
+**API Version Errors**
+- Try updating `AZURE_OPENAI_API_VERSION` to a supported version
+- Check [Azure OpenAI API versions](https://learn.microsoft.com/en-us/azure/ai-services/openai/reference) for the latest
+
+### Verifying Configuration
+
+List available models to verify your configuration:
+
+```bash
+moltbot models list
+```
+
+The Azure OpenAI deployment should appear as `azure-openai/{deployment-name}`.
+
+## Security Considerations
+
+- Store API keys securely using environment variables or a secrets manager
+- Consider using Azure Managed Identity for production deployments
+- Review Azure OpenAI content filtering policies for your use case
+- Ensure your Azure resource has appropriate network access controls
+
+## Related Documentation
+
+- [Azure OpenAI Service Documentation](https://learn.microsoft.com/en-us/azure/ai-services/openai/)
+- [Azure OpenAI API Reference](https://learn.microsoft.com/en-us/azure/ai-services/openai/reference)
+- [Moltbot Configuration](/configuration)
--- a/src/agents/azure-openai-provider.ts
+++ b/src/agents/azure-openai-provider.ts
@ -0,0 +1,342 @@
+/**
+ * Azure OpenAI Provider Configuration
+ *
+ * This module provides support for Azure OpenAI Service, which uses a different
+ * API endpoint format and authentication mechanism compared to OpenAI's standard API.
+ *
+ * Azure OpenAI endpoint format:
+ * https://{resourceName}.openai.azure.com/openai/deployments/{deploymentName}/chat/completions?api-version={apiVersion}
+ */
+
+import type { ModelDefinitionConfig } from "../config/types.models.js";
+import type { ProviderConfig } from "./models-config.providers.js";
+
+// Store for Azure OpenAI configurations to enable fetch interception
+const azureOpenAIConfigs = new Map<string, { apiVersion: string }>();
+
+/**
+ * Register an Azure OpenAI resource for fetch interception.
+ * This enables the global fetch wrapper to add api-version query parameters.
+ */
+export function registerAzureOpenAIResource(resourceName: string, apiVersion: string): void {
+  azureOpenAIConfigs.set(resourceName.toLowerCase(), { apiVersion });
+}
+
+/**
+ * Check if a URL is an Azure OpenAI endpoint and get its configuration.
+ */
+export function getAzureOpenAIConfig(
+  url: string,
+): { resourceName: string; apiVersion: string } | null {
+  try {
+    const parsed = new URL(url);
+    const match = /^([^.]+)\.openai\.azure\.com$/.exec(parsed.hostname);
+    if (!match) return null;
+    const resourceName = match[1].toLowerCase();
+    const config = azureOpenAIConfigs.get(resourceName);
+    if (!config) return null;
+    return { resourceName, apiVersion: config.apiVersion };
+  } catch {
+    return null;
+  }
+}
+
+// Track if the fetch wrapper has been installed
+let fetchWrapperInstalled = false;
+let originalFetch: typeof fetch | null = null;
+
+/**
+ * Install a global fetch wrapper that adds api-version query parameter
+ * to Azure OpenAI requests.
+ */
+export function installAzureOpenAIFetchWrapper(): void {
+  if (fetchWrapperInstalled) return;
+
+  originalFetch = globalThis.fetch;
+  fetchWrapperInstalled = true;
+
+  globalThis.fetch = async function azureOpenAIFetchWrapper(
+    input: RequestInfo | URL,
+    init?: RequestInit,
+  ): Promise<Response> {
+    const url =
+      typeof input === "string" ? input : input instanceof URL ? input.toString() : input.url;
+    const azureConfig = getAzureOpenAIConfig(url);
+
+    if (azureConfig) {
+      // Add api-version query parameter to Azure OpenAI requests
+      const parsedUrl = new URL(url);
+      if (!parsedUrl.searchParams.has("api-version")) {
+        parsedUrl.searchParams.set("api-version", azureConfig.apiVersion);
+        const newUrl = parsedUrl.toString();
+
+        // Recreate the request with the new URL
+        if (typeof input === "string") {
+          return originalFetch!(newUrl, init);
+        } else if (input instanceof URL) {
+          return originalFetch!(new URL(newUrl), init);
+        } else {
+          // Request object - create new request with modified URL
+          const newRequest = new Request(newUrl, {
+            method: input.method,
+            headers: input.headers,
+            body: init?.body ?? input.body,
+            mode: input.mode,
+            credentials: input.credentials,
+            cache: input.cache,
+            redirect: input.redirect,
+            referrer: input.referrer,
+            integrity: input.integrity,
+            signal: init?.signal ?? input.signal,
+          });
+          return originalFetch!(newRequest);
+        }
+      }
+    }
+
+    return originalFetch!(input, init);
+  };
+}
+
+/**
+ * Uninstall the Azure OpenAI fetch wrapper (for testing).
+ */
+export function uninstallAzureOpenAIFetchWrapper(): void {
+  if (!fetchWrapperInstalled || !originalFetch) return;
+  globalThis.fetch = originalFetch;
+  fetchWrapperInstalled = false;
+  originalFetch = null;
+}
+
+// Azure OpenAI default configuration
+const AZURE_OPENAI_DEFAULT_API_VERSION = "2024-08-01-preview";
+const AZURE_OPENAI_DEFAULT_CONTEXT_WINDOW = 128000;
+const AZURE_OPENAI_DEFAULT_MAX_TOKENS = 4096;
+const AZURE_OPENAI_DEFAULT_COST = {
+  input: 0,
+  output: 0,
+  cacheRead: 0,
+  cacheWrite: 0,
+};
+
+export interface AzureOpenAIConfig {
+  /** Azure OpenAI resource name (the name of your Azure OpenAI resource) */
+  resourceName: string;
+  /** Deployment name (the name of your model deployment) */
+  deploymentName: string;
+  /** API version (defaults to 2024-08-01-preview) */
+  apiVersion?: string;
+  /** Azure OpenAI API key */
+  apiKey?: string;
+  /** Custom model display name */
+  modelName?: string;
+  /** Whether this is a reasoning model (like o1) */
+  reasoning?: boolean;
+  /** Supported input types */
+  input?: Array<"text" | "image">;
+  /** Context window size */
+  contextWindow?: number;
+  /** Max output tokens */
+  maxTokens?: number;
+}
+
+/**
+ * Builds the Azure OpenAI base URL from resource name
+ */
+export function buildAzureOpenAIBaseUrl(resourceName: string): string {
+  return `https://${resourceName}.openai.azure.com`;
+}
+
+/**
+ * Builds the full Azure OpenAI API endpoint URL
+ */
+export function buildAzureOpenAIEndpoint(
+  resourceName: string,
+  deploymentName: string,
+  apiVersion: string = AZURE_OPENAI_DEFAULT_API_VERSION,
+): string {
+  return `${buildAzureOpenAIBaseUrl(resourceName)}/openai/deployments/${deploymentName}/chat/completions?api-version=${apiVersion}`;
+}
+
+/**
+ * Creates a model definition for Azure OpenAI deployment
+ */
+export function buildAzureOpenAIModelDefinition(config: AzureOpenAIConfig): ModelDefinitionConfig {
+  const modelId = config.deploymentName;
+  const modelName = config.modelName ?? `Azure ${config.deploymentName}`;
+
+  return {
+    id: modelId,
+    name: modelName,
+    reasoning: config.reasoning ?? false,
+    input: config.input ?? ["text"],
+    cost: AZURE_OPENAI_DEFAULT_COST,
+    contextWindow: config.contextWindow ?? AZURE_OPENAI_DEFAULT_CONTEXT_WINDOW,
+    maxTokens: config.maxTokens ?? AZURE_OPENAI_DEFAULT_MAX_TOKENS,
+  };
+}
+
+/**
+ * Builds a provider configuration for Azure OpenAI
+ *
+ * Azure OpenAI uses a different URL format:
+ * https://{resource}.openai.azure.com/openai/deployments/{deployment}/chat/completions?api-version={version}
+ *
+ * We set the baseUrl to include the deployment path so the OpenAI SDK appends /chat/completions correctly.
+ * The api-version query parameter is added automatically via a global fetch wrapper.
+ */
+export function buildAzureOpenAIProvider(config: AzureOpenAIConfig): ProviderConfig {
+  const apiVersion = config.apiVersion ?? AZURE_OPENAI_DEFAULT_API_VERSION;
+  // Azure OpenAI requires the deployment in the URL path
+  // Format: https://{resource}.openai.azure.com/openai/deployments/{deployment}
+  // The SDK will append /chat/completions to this
+  const baseUrl = `https://${config.resourceName}.openai.azure.com/openai/deployments/${config.deploymentName}`;
+
+  // Register the resource for fetch interception to add api-version query param
+  registerAzureOpenAIResource(config.resourceName, apiVersion);
+  // Install the global fetch wrapper
+  installAzureOpenAIFetchWrapper();
+
+  return {
+    baseUrl,
+    api: "openai-completions",
+    apiKey: config.apiKey,
+    // Azure OpenAI uses api-key header instead of Bearer token
+    authHeader: false,
+    headers: {
+      "api-key": config.apiKey ?? "",
+    },
+    // Azure requires api-version as query parameter - store in provider config
+    azureResourceName: config.resourceName,
+    azureApiVersion: apiVersion,
+    models: [buildAzureOpenAIModelDefinition(config)],
+  };
+}
+
+/**
+ * Builds a provider configuration for Azure OpenAI with multiple deployments
+ *
+ * Note: Each deployment gets its own baseUrl with the deployment name in the path.
+ * For multiple deployments, we use the first deployment in the baseUrl.
+ * The api-version query parameter is added automatically via a global fetch wrapper.
+ */
+export function buildAzureOpenAIProviderMultiDeployment(params: {
+  resourceName: string;
+  apiKey?: string;
+  apiVersion?: string;
+  deployments: Array<{
+    name: string;
+    displayName?: string;
+    reasoning?: boolean;
+    input?: Array<"text" | "image">;
+    contextWindow?: number;
+    maxTokens?: number;
+  }>;
+}): ProviderConfig {
+  const apiVersion = params.apiVersion ?? AZURE_OPENAI_DEFAULT_API_VERSION;
+  const firstDeployment = params.deployments[0]?.name ?? "default";
+  const baseUrl = `https://${params.resourceName}.openai.azure.com/openai/deployments/${firstDeployment}`;
+
+  // Register the resource for fetch interception to add api-version query param
+  registerAzureOpenAIResource(params.resourceName, apiVersion);
+  // Install the global fetch wrapper
+  installAzureOpenAIFetchWrapper();
+
+  const models: ModelDefinitionConfig[] = params.deployments.map((deployment) => ({
+    id: deployment.name,
+    name: deployment.displayName ?? `Azure ${deployment.name}`,
+    reasoning: deployment.reasoning ?? false,
+    input: deployment.input ?? ["text"],
+    cost: AZURE_OPENAI_DEFAULT_COST,
+    contextWindow: deployment.contextWindow ?? AZURE_OPENAI_DEFAULT_CONTEXT_WINDOW,
+    maxTokens: deployment.maxTokens ?? AZURE_OPENAI_DEFAULT_MAX_TOKENS,
+  }));
+
+  return {
+    baseUrl,
+    api: "openai-completions",
+    apiKey: params.apiKey,
+    authHeader: false,
+    headers: {
+      "api-key": params.apiKey ?? "",
+    },
+    azureResourceName: params.resourceName,
+    azureApiVersion: apiVersion,
+    models,
+  };
+}
+
+/**
+ * Environment variable names for Azure OpenAI configuration
+ */
+export const AZURE_OPENAI_ENV = {
+  API_KEY: "AZURE_OPENAI_API_KEY",
+  RESOURCE_NAME: "AZURE_OPENAI_RESOURCE_NAME",
+  DEPLOYMENT_NAME: "AZURE_OPENAI_DEPLOYMENT_NAME",
+  API_VERSION: "AZURE_OPENAI_API_VERSION",
+  ENDPOINT: "AZURE_OPENAI_ENDPOINT",
+} as const;
+
+/**
+ * Resolves Azure OpenAI configuration from environment variables
+ */
+export function resolveAzureOpenAIConfigFromEnv(
+  env: NodeJS.ProcessEnv = process.env,
+): AzureOpenAIConfig | null {
+  const apiKey = env[AZURE_OPENAI_ENV.API_KEY]?.trim();
+  const resourceName = env[AZURE_OPENAI_ENV.RESOURCE_NAME]?.trim();
+  const deploymentName = env[AZURE_OPENAI_ENV.DEPLOYMENT_NAME]?.trim();
+  const apiVersion = env[AZURE_OPENAI_ENV.API_VERSION]?.trim();
+
+  // If endpoint is provided directly, parse resource name from it
+  const endpoint = env[AZURE_OPENAI_ENV.ENDPOINT]?.trim();
+  let resolvedResourceName = resourceName;
+  if (!resolvedResourceName && endpoint) {
+    const match = /https:\/\/([^.]+)\.openai\.azure\.com/.exec(endpoint);
+    if (match) {
+      resolvedResourceName = match[1];
+    }
+  }
+
+  if (!apiKey || !resolvedResourceName || !deploymentName) {
+    return null;
+  }
+
+  return {
+    resourceName: resolvedResourceName,
+    deploymentName,
+    apiVersion: apiVersion || AZURE_OPENAI_DEFAULT_API_VERSION,
+    apiKey,
+  };
+}
+
+/**
+ * Default export for common models available on Azure OpenAI
+ * These are the most commonly deployed models
+ */
+export const AZURE_OPENAI_COMMON_MODELS = {
+  "gpt-5": {
+    reasoning: false,
+    input: ["text", "image"] as Array<"text" | "image">,
+    contextWindow: AZURE_OPENAI_DEFAULT_CONTEXT_WINDOW,
+    maxTokens: AZURE_OPENAI_DEFAULT_MAX_TOKENS,
+  },
+  "gpt-5-mini": {
+    reasoning: false,
+    input: ["text", "image"] as Array<"text" | "image">,
+    contextWindow: AZURE_OPENAI_DEFAULT_CONTEXT_WINDOW,
+    maxTokens: AZURE_OPENAI_DEFAULT_MAX_TOKENS,
+  },
+  "gpt-5-nano": {
+    reasoning: false,
+    input: ["text", "image"] as Array<"text" | "image">,
+    contextWindow: AZURE_OPENAI_DEFAULT_CONTEXT_WINDOW,
+    maxTokens: AZURE_OPENAI_DEFAULT_MAX_TOKENS,
+  },
+  "gpt-5-codex": {
+    reasoning: false,
+    input: ["text"] as Array<"text" | "image">,
+    contextWindow: AZURE_OPENAI_DEFAULT_CONTEXT_WINDOW,
+    maxTokens: AZURE_OPENAI_DEFAULT_MAX_TOKENS,
+  },
+} as const;
--- a/src/agents/model-auth.ts
+++ b/src/agents/model-auth.ts
@ -286,6 +286,7 @@ export function resolveEnvApiKey(provider: string): EnvApiKeyResult | null {
    venice: "VENICE_API_KEY",
    mistral: "MISTRAL_API_KEY",
    opencode: "OPENCODE_API_KEY",
+    "azure-openai": "AZURE_OPENAI_API_KEY",
  };
  const envVar = envMap[normalized];
  if (!envVar) return null;
--- a/src/agents/models-config.providers.ts
+++ b/src/agents/models-config.providers.ts
@ -13,6 +13,11 @@ import {
  SYNTHETIC_MODEL_CATALOG,
 } from "./synthetic-models.js";
 import { discoverVeniceModels, VENICE_BASE_URL } from "./venice-models.js";
+import {
+  buildAzureOpenAIProvider,
+  resolveAzureOpenAIConfigFromEnv,
+  AZURE_OPENAI_ENV,
+} from "./azure-openai-provider.js";

 type ModelsConfig = NonNullable<OpenClawConfig["models"]>;
 export type ProviderConfig = NonNullable<ModelsConfig["providers"]>[string];
@ -454,6 +459,30 @@ export async function resolveImplicitProviders(params: {
    providers.ollama = { ...(await buildOllamaProvider()), apiKey: ollamaKey };
  }

+  // Azure OpenAI provider - auto-discover from environment variables
+  const azureOpenAIConfig = resolveAzureOpenAIConfigFromEnv();
+  if (azureOpenAIConfig) {
+    providers["azure-openai"] = buildAzureOpenAIProvider(azureOpenAIConfig);
+  } else {
+    // Check for API key in auth profiles
+    const azureOpenAIKey =
+      resolveEnvApiKeyVarName("azure-openai") ??
+      resolveApiKeyFromProfiles({ provider: "azure-openai", store: authStore });
+    if (azureOpenAIKey) {
+      // If we have an API key but not full config, check for resource/deployment in env
+      const resourceName = process.env[AZURE_OPENAI_ENV.RESOURCE_NAME]?.trim();
+      const deploymentName = process.env[AZURE_OPENAI_ENV.DEPLOYMENT_NAME]?.trim();
+      if (resourceName && deploymentName) {
+        providers["azure-openai"] = buildAzureOpenAIProvider({
+          resourceName,
+          deploymentName,
+          apiKey: azureOpenAIKey,
+          apiVersion: process.env[AZURE_OPENAI_ENV.API_VERSION]?.trim(),
+        });
+      }
+    }
+  }
+
  return providers;
 }

--- a/src/config/types.models.ts
+++ b/src/config/types.models.ts
@ -41,6 +41,10 @@ export type ModelProviderConfig = {
  headers?: Record<string, string>;
  authHeader?: boolean;
  models: ModelDefinitionConfig[];
+  /** Azure OpenAI specific: resource name */
+  azureResourceName?: string;
+  /** Azure OpenAI specific: API version */
+  azureApiVersion?: string;
 };

 export type BedrockDiscoveryConfig = {