genai-lite 0.8.3 → 0.9.0

package/LICENSE

@@ -1,21 +1,21 @@
-MIT License
-
-Copyright (c) 2025 Luigi Acerbi
-
-Permission is hereby granted, free of charge, to any person obtaining a copy
-of this software and associated documentation files (the "Software"), to deal
-in the Software without restriction, including without limitation the rights
-to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
-copies of the Software, and to permit persons to whom the Software is
-furnished to do so, subject to the following conditions:
-
-The above copyright notice and this permission notice shall be included in all
-copies or substantial portions of the Software.
-
-THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
-IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
-FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
-AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
-LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
-OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
-SOFTWARE.
+MIT License
+
+Copyright (c) 2025 Luigi Acerbi
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/README.md

@@ -1,207 +1,210 @@
-# genai-lite
-
-A lightweight, portable Node.js/TypeScript library providing a unified interface for interacting with multiple Generative AI providers—both cloud-based (OpenAI, Anthropic, Google Gemini, Mistral) and local (llama.cpp, stable-diffusion.cpp). Supports both LLM chat and AI image generation.
-
-## Features
-
-- 🔌 **Unified API** - Single interface for multiple AI providers
-- 🏠 **Local & Cloud Models** - Run models locally with llama.cpp or use cloud APIs
-- 🖼️ **Image Generation** - First-class support for AI image generation (OpenAI, local diffusion)
-- 🔐 **Flexible API Key Management** - Bring your own key storage solution
-- 📦 **Zero Electron Dependencies** - Works in any Node.js environment
-- 🎯 **TypeScript First** - Full type safety and IntelliSense support
-- ⚡ **Lightweight** - Minimal dependencies, focused functionality
-- 🛡️ **Provider Normalization** - Consistent responses across different AI APIs
-- 🎨 **Configurable Model Presets** - Built-in presets with full customization options
-- 🎭 **Template Engine** - Sophisticated templating with conditionals and variable substitution
-- 📊 **Configurable Logging** - Debug mode, custom loggers (pino, winston), and silent mode for tests
-
-## Installation
-
-```bash
-npm install genai-lite
-```
-
-Set API keys as environment variables:
-
-```bash
-export OPENAI_API_KEY=sk-...
-export ANTHROPIC_API_KEY=sk-ant-...
-export GEMINI_API_KEY=AIza...
-```
-
-## Quick Start
-
-### Cloud Providers (OpenAI, Anthropic, Gemini, Mistral)
-
-```typescript
-import { LLMService, fromEnvironment } from 'genai-lite';
-
-const llmService = new LLMService(fromEnvironment);
-
-const response = await llmService.sendMessage({
-  providerId: 'openai',
-  modelId: 'gpt-4.1-mini',
-  messages: [
-    { role: 'system', content: 'You are a helpful assistant.' },
-    { role: 'user', content: 'Hello, how are you?' }
-  ]
-});
-
-if (response.object === 'chat.completion') {
-  console.log(response.choices[0].message.content);
-}
-```
-
-### Local Models (llama.cpp)
-
-```typescript
-import { LLMService } from 'genai-lite';
-
-// Start llama.cpp server first: llama-server -m /path/to/model.gguf --port 8080
-const llmService = new LLMService(async () => 'not-needed');
-
-const response = await llmService.sendMessage({
-  providerId: 'llamacpp',
-  modelId: 'llamacpp',  // Generic ID for whatever model is loaded
-  messages: [
-    { role: 'system', content: 'You are a helpful assistant.' },
-    { role: 'user', content: 'Explain quantum computing briefly.' }
-  ]
-});
-
-if (response.object === 'chat.completion') {
-  console.log(response.choices[0].message.content);
-}
-```
-
-### Image Generation
-
-```typescript
-import { ImageService, fromEnvironment } from 'genai-lite';
-
-const imageService = new ImageService(fromEnvironment);
-
-const result = await imageService.generateImage({
-  providerId: 'openai-images',
-  modelId: 'gpt-image-1-mini',
-  prompt: 'A serene mountain lake at sunrise, photorealistic',
-  settings: {
-    width: 1024,
-    height: 1024,
-    quality: 'high'
-  }
-});
-
-if (result.object === 'image.result') {
-  require('fs').writeFileSync('output.png', result.data[0].data);
-}
-```
-
-## Documentation
-
-Comprehensive documentation is available in the **[`genai-lite-docs`](./genai-lite-docs/index.md)** folder.
-
-### Getting Started
-- **[Documentation Hub](./genai-lite-docs/index.md)** - Navigation and overview
-- **[Core Concepts](./genai-lite-docs/core-concepts.md)** - API keys, presets, settings, errors
-
-### API Reference
-- **[LLM Service](./genai-lite-docs/llm-service.md)** - Text generation and chat
-- **[Image Service](./genai-lite-docs/image-service.md)** - Image generation (cloud and local)
-- **[llama.cpp Integration](./genai-lite-docs/llamacpp-integration.md)** - Local LLM inference
-
-### Utilities & Advanced
-- **[Prompting Utilities](./genai-lite-docs/prompting-utilities.md)** - Template engine, token counting, content parsing
-- **[Logging](./genai-lite-docs/logging.md)** - Configure logging and debugging
-- **[TypeScript Reference](./genai-lite-docs/typescript-reference.md)** - Type definitions
-
-### Provider Reference
-- **[Providers & Models](./genai-lite-docs/providers-and-models.md)** - Supported providers and models
-
-### Examples & Help
-- **[Example: Chat Demo](./genai-lite-docs/example-chat-demo.md)** - Reference implementation for chat applications
-- **[Example: Image Demo](./genai-lite-docs/example-image-demo.md)** - Reference implementation for image generation applications
-- **[Troubleshooting](./genai-lite-docs/troubleshooting.md)** - Common issues and solutions
-
-## Supported Providers
-
-### LLM Providers
-- **OpenAI** - GPT-5 (5.2, 5.1, mini, nano), GPT-4.1, o4-mini
-- **Anthropic** - Claude 4.5 (Opus, Sonnet, Haiku), Claude 4, Claude 3.7, Claude 3.5
-- **Google Gemini** - Gemini 3 (Pro, Flash preview), Gemini 2.5, Gemini 2.0, Gemma 3 (free)
-- **Mistral** - Codestral, Devstral
-- **llama.cpp** - Run any GGUF model locally (no API keys required)
-
-### Image Providers
-- **OpenAI Images** - gpt-image-1, dall-e-3, dall-e-2
-- **genai-electron** - Local Stable Diffusion models
-
-See **[Providers & Models](./genai-lite-docs/providers-and-models.md)** for complete model listings and capabilities.
-
-## API Key Management
-
-genai-lite uses a flexible API key provider pattern. Use the built-in `fromEnvironment` provider or create your own:
-
-```typescript
-import { ApiKeyProvider, LLMService } from 'genai-lite';
-
-const myKeyProvider: ApiKeyProvider = async (providerId: string) => {
-  const key = await mySecureStorage.getKey(providerId);
-  return key || null;
-};
-
-const llmService = new LLMService(myKeyProvider);
-```
-
-See **[Core Concepts](./genai-lite-docs/core-concepts.md#api-key-management)** for detailed examples including Electron integration.
-
-## Logging Configuration
-
-Control logging verbosity via environment variable or service options:
-
-```bash
-# Environment variable (applies to all services)
-export GENAI_LITE_LOG_LEVEL=debug  # Options: silent, error, warn, info, debug
-```
-
-```typescript
-// Per-service configuration
-const llmService = new LLMService(fromEnvironment, {
-  logLevel: 'debug',        // Override env var
-  logger: customPinoLogger  // Inject pino/winston/etc.
-});
-```
-
-See **[Logging](./genai-lite-docs/logging.md)** for custom logger integration and testing patterns.
-
-## Example Applications
-
-The library includes two complete demo applications showcasing all features:
-
-- **[chat-demo](examples/chat-demo)** - Interactive chat application with all LLM providers, template rendering, and advanced features
-- **[image-gen-demo](examples/image-gen-demo)** - Interactive image generation UI with OpenAI and local diffusion support
-
-Both demos are production-ready React + Express applications that serve as reference implementations and testing environments. See **[Example: Chat Demo](./genai-lite-docs/example-chat-demo.md)** and **[Example: Image Demo](./genai-lite-docs/example-image-demo.md)** for detailed documentation.
-
-## Contributing
-
-Contributions are welcome! Please feel free to submit a Pull Request. For major changes, please open an issue first to discuss what you would like to change.
-
-### Development
-
-```bash
-npm install
-npm run build
-npm test
-```
-
-See **[Troubleshooting](./genai-lite-docs/troubleshooting.md)** for information about E2E tests and development workflows.
-
-## License
-
-This project is licensed under the MIT License - see the LICENSE file for details.
-
-## Acknowledgments
-
-Originally developed as part of the Athanor project, genai-lite has been extracted and made standalone to benefit the wider developer community.
+# genai-lite
+
+A lightweight, portable Node.js/TypeScript library providing a unified interface for interacting with multiple Generative AI providers—both cloud-based (OpenAI, Anthropic, Google Gemini, Mistral, OpenRouter) and local (llama.cpp, stable-diffusion.cpp). Supports both LLM chat and AI image generation.
+
+## Features
+
+- 🔌 **Unified API** - Single interface for multiple AI providers
+- 🏠 **Local & Cloud Models** - Run models locally with llama.cpp or use cloud APIs
+- 🖼️ **Image Generation** - First-class support for AI image generation (OpenAI, local diffusion)
+- 🔐 **Flexible API Key Management** - Bring your own key storage solution
+- 📦 **Zero Electron Dependencies** - Works in any Node.js environment
+- 🎯 **TypeScript First** - Full type safety and IntelliSense support
+- ⚡ **Lightweight** - Minimal dependencies, focused functionality
+- 🛡️ **Provider Normalization** - Consistent responses across different AI APIs
+- 🧠 **Local Reasoning Toggle** - Turn thinking on/off for detected GGUF models (Qwen 3.5-class, Gemma 4, and more) with vendor-tuned sampling defaults
+- 🔁 **Built-in Reliability** - Automatic retries with backoff/`Retry-After`, per-request timeouts, and abort signals
+- 🎨 **Configurable Model Presets** - Built-in presets with full customization options
+- 🎭 **Template Engine** - Sophisticated templating with conditionals and variable substitution
+- 📊 **Configurable Logging** - Debug mode, custom loggers (pino, winston), and silent mode for tests
+
+## Installation
+
+```bash
+npm install genai-lite
+```
+
+Set API keys as environment variables:
+
+```bash
+export OPENAI_API_KEY=sk-...
+export ANTHROPIC_API_KEY=sk-ant-...
+export GEMINI_API_KEY=AIza...
+```
+
+## Quick Start
+
+### Cloud Providers (OpenAI, Anthropic, Gemini, Mistral)
+
+```typescript
+import { LLMService, fromEnvironment } from 'genai-lite';
+
+const llmService = new LLMService(fromEnvironment);
+
+const response = await llmService.sendMessage({
+  providerId: 'openai',
+  modelId: 'gpt-4.1-mini',
+  messages: [
+    { role: 'system', content: 'You are a helpful assistant.' },
+    { role: 'user', content: 'Hello, how are you?' }
+  ]
+});
+
+if (response.object === 'chat.completion') {
+  console.log(response.choices[0].message.content);
+}
+```
+
+### Local Models (llama.cpp)
+
+```typescript
+import { LLMService } from 'genai-lite';
+
+// Start llama.cpp server first: llama-server -m /path/to/model.gguf --port 8080
+const llmService = new LLMService(async () => 'not-needed');
+
+const response = await llmService.sendMessage({
+  providerId: 'llamacpp',
+  modelId: 'llamacpp',  // Generic ID for whatever model is loaded
+  messages: [
+    { role: 'system', content: 'You are a helpful assistant.' },
+    { role: 'user', content: 'Explain quantum computing briefly.' }
+  ]
+});
+
+if (response.object === 'chat.completion') {
+  console.log(response.choices[0].message.content);
+}
+```
+
+### Image Generation
+
+```typescript
+import { ImageService, fromEnvironment } from 'genai-lite';
+
+const imageService = new ImageService(fromEnvironment);
+
+const result = await imageService.generateImage({
+  providerId: 'openai-images',
+  modelId: 'gpt-image-1-mini',
+  prompt: 'A serene mountain lake at sunrise, photorealistic',
+  settings: {
+    width: 1024,
+    height: 1024,
+    quality: 'high'
+  }
+});
+
+if (result.object === 'image.result') {
+  require('fs').writeFileSync('output.png', result.data[0].data);
+}
+```
+
+## Documentation
+
+Comprehensive documentation is available in the **[`genai-lite-docs`](./genai-lite-docs/index.md)** folder.
+
+### Getting Started
+- **[Documentation Hub](./genai-lite-docs/index.md)** - Navigation and overview
+- **[Core Concepts](./genai-lite-docs/core-concepts.md)** - API keys, presets, settings, errors
+
+### API Reference
+- **[LLM Service](./genai-lite-docs/llm-service.md)** - Text generation and chat
+- **[Image Service](./genai-lite-docs/image-service.md)** - Image generation (cloud and local)
+- **[llama.cpp Integration](./genai-lite-docs/llamacpp-integration.md)** - Local LLM inference
+
+### Utilities & Advanced
+- **[Prompting Utilities](./genai-lite-docs/prompting-utilities.md)** - Template engine, token counting, content parsing
+- **[Logging](./genai-lite-docs/logging.md)** - Configure logging and debugging
+- **[TypeScript Reference](./genai-lite-docs/typescript-reference.md)** - Type definitions
+
+### Provider Reference
+- **[Providers & Models](./genai-lite-docs/providers-and-models.md)** - Supported providers and models
+
+### Examples & Help
+- **[Example: Chat Demo](./genai-lite-docs/example-chat-demo.md)** - Reference implementation for chat applications
+- **[Example: Image Demo](./genai-lite-docs/example-image-demo.md)** - Reference implementation for image generation applications
+- **[Troubleshooting](./genai-lite-docs/troubleshooting.md)** - Common issues and solutions
+
+## Supported Providers
+
+### LLM Providers
+- **OpenAI** - GPT-5 (5.2, 5.1, mini, nano), GPT-4.1, o4-mini
+- **Anthropic** - Claude 4.5 (Opus, Sonnet, Haiku), Claude 4, Claude 3.7, Claude 3.5
+- **Google Gemini** - Gemini 3 (Pro, Flash preview), Gemini 2.5, Gemma 3 & 4 (free)
+- **Mistral** - Codestral, Devstral
+- **OpenRouter** - Unified gateway to 100+ models (unknown models assumed reasoning-capable)
+- **llama.cpp** - Run any GGUF model locally (no API keys required); local reasoning toggle for detected models
+
+### Image Providers
+- **OpenAI Images** - gpt-image-1, dall-e-3, dall-e-2
+- **genai-electron** - Local Stable Diffusion models
+
+See **[Providers & Models](./genai-lite-docs/providers-and-models.md)** for complete model listings and capabilities.
+
+## API Key Management
+
+genai-lite uses a flexible API key provider pattern. Use the built-in `fromEnvironment` provider or create your own:
+
+```typescript
+import { ApiKeyProvider, LLMService } from 'genai-lite';
+
+const myKeyProvider: ApiKeyProvider = async (providerId: string) => {
+  const key = await mySecureStorage.getKey(providerId);
+  return key || null;
+};
+
+const llmService = new LLMService(myKeyProvider);
+```
+
+See **[Core Concepts](./genai-lite-docs/core-concepts.md#api-key-management)** for detailed examples including Electron integration.
+
+## Logging Configuration
+
+Control logging verbosity via environment variable or service options:
+
+```bash
+# Environment variable (applies to all services)
+export GENAI_LITE_LOG_LEVEL=debug  # Options: silent, error, warn, info, debug
+```
+
+```typescript
+// Per-service configuration
+const llmService = new LLMService(fromEnvironment, {
+  logLevel: 'debug',        // Override env var
+  logger: customPinoLogger  // Inject pino/winston/etc.
+});
+```
+
+See **[Logging](./genai-lite-docs/logging.md)** for custom logger integration and testing patterns.
+
+## Example Applications
+
+The library includes two complete demo applications showcasing all features:
+
+- **[chat-demo](examples/chat-demo)** - Interactive chat application with all LLM providers, template rendering, and advanced features
+- **[image-gen-demo](examples/image-gen-demo)** - Interactive image generation UI with OpenAI and local diffusion support
+
+Both demos are production-ready React + Express applications that serve as reference implementations and testing environments. See **[Example: Chat Demo](./genai-lite-docs/example-chat-demo.md)** and **[Example: Image Demo](./genai-lite-docs/example-image-demo.md)** for detailed documentation.
+
+## Contributing
+
+Contributions are welcome! Please feel free to submit a Pull Request. For major changes, please open an issue first to discuss what you would like to change.
+
+### Development
+
+```bash
+npm install
+npm run build
+npm test
+```
+
+See **[Troubleshooting](./genai-lite-docs/troubleshooting.md)** for information about E2E tests and development workflows.
+
+## License
+
+This project is licensed under the MIT License - see the LICENSE file for details.
+
+## Acknowledgments
+
+Originally developed as part of the Athanor project, genai-lite has been extracted and made standalone to benefit the wider developer community.

package/dist/config/llm-presets.json

@@ -1,4 +1,28 @@
 [
+    {
+        "id": "llamacpp-local-default",
+        "displayName": "llama.cpp - Local Model",
+        "description": "Local GGUF model via llama.cpp server. Vendor sampling defaults and thinking-off are applied automatically for recognized models (Qwen 3.x, Gemma 3/4, GPT-OSS, Ministral 3, Granite 4.1, ...).",
+        "providerId": "llamacpp",
+        "modelId": "llamacpp",
+        "settings": {
+            "reasoning": {
+                "enabled": false
+            }
+        }
+    },
+    {
+        "id": "llamacpp-local-thinking",
+        "displayName": "llama.cpp - Local Model (Thinking)",
+        "description": "Local GGUF model with reasoning/thinking enabled. For hybrid models (Qwen 3.5/3.6, Gemma 4) this sends enable_thinking=true; the reasoning trace is returned on choice.reasoning.",
+        "providerId": "llamacpp",
+        "modelId": "llamacpp",
+        "settings": {
+            "reasoning": {
+                "enabled": true
+            }
+        }
+    },
     {
         "id": "anthropic-claude-opus-4-5-20251101-default",
         "displayName": "Anthropic - Claude Opus 4.5",

package/dist/index.d.ts

@@ -1,6 +1,8 @@
 export type { ApiKeyProvider, PresetMode } from "./types";
 export { LLMService } from "./llm/LLMService";
-export type { LLMServiceOptions, CreateMessagesResult } from "./llm/LLMService";
+export type { LLMServiceOptions, SendMessageOptions, CreateMessagesResult } from "./llm/LLMService";
+export { withRetry, DEFAULT_RETRY_POLICY } from "./shared/services/withRetry";
+export type { RetryPolicy, RetryVerdict, WithRetryOptions } from "./shared/services/withRetry";
 export type { ModelPreset } from "./types/presets";
 export * from "./llm/types";
 export * from "./llm/clients/types";
@@ -17,7 +19,7 @@ export { ImageService } from "./image/ImageService";
 export type { ImageProviderId, ImageMimeType, ImageResponseFormat, ImageQuality, ImageStyle, DiffusionSampler, ImageProgressStage, ImageProgressCallback, DiffusionSettings, OpenAISpecificSettings, ImageGenerationSettings, ResolvedImageGenerationSettings, ImageUsage, GeneratedImage, ImageGenerationRequestBase, ImageGenerationRequest, ImageGenerationRequestWithPreset, ImageGenerationResponse, ImageFailureResponse, ImageProviderCapabilities, ImageModelInfo, ImageProviderInfo, ImagePreset, ImageProviderAdapterConfig, ImageProviderAdapter, ImageServiceOptions, CreatePromptResult, } from "./types/image";
 export { renderTemplate } from "./prompting/template";
 export { countTokens, getSmartPreview, extractRandomVariables } from "./prompting/content";
-export { parseStructuredContent, parseRoleTags, extractInitialTaggedContent, parseTemplateWithMetadata } from "./prompting/parser";
+export { parseStructuredContent, parseRoleTags, extractInitialTaggedContent, extractMarkerDelimitedContent, parseTemplateWithMetadata } from "./prompting/parser";
 export type { TemplateMetadata } from "./prompting/parser";
 export { createFallbackModelInfo, detectGgufCapabilities, KNOWN_GGUF_MODELS } from "./llm/config";
 export type { GgufModelPattern } from "./llm/config";

package/dist/index.js

@@ -14,10 +14,14 @@ var __exportStar = (this && this.__exportStar) || function(m, exports) {
     for (var p in m) if (p !== "default" && !Object.prototype.hasOwnProperty.call(exports, p)) __createBinding(exports, m, p);
 };
 Object.defineProperty(exports, "__esModule", { value: true });
-exports.silentLogger = exports.DEFAULT_LOG_LEVEL = exports.createDefaultLogger = exports.KNOWN_GGUF_MODELS = exports.detectGgufCapabilities = exports.createFallbackModelInfo = exports.parseTemplateWithMetadata = exports.extractInitialTaggedContent = exports.parseRoleTags = exports.parseStructuredContent = exports.extractRandomVariables = exports.getSmartPreview = exports.countTokens = exports.renderTemplate = exports.ImageService = exports.MistralClientAdapter = exports.OpenRouterClientAdapter = exports.LlamaCppServerClient = exports.LlamaCppClientAdapter = exports.fromEnvironment = exports.LLMService = void 0;
+exports.silentLogger = exports.DEFAULT_LOG_LEVEL = exports.createDefaultLogger = exports.KNOWN_GGUF_MODELS = exports.detectGgufCapabilities = exports.createFallbackModelInfo = exports.parseTemplateWithMetadata = exports.extractMarkerDelimitedContent = exports.extractInitialTaggedContent = exports.parseRoleTags = exports.parseStructuredContent = exports.extractRandomVariables = exports.getSmartPreview = exports.countTokens = exports.renderTemplate = exports.ImageService = exports.MistralClientAdapter = exports.OpenRouterClientAdapter = exports.LlamaCppServerClient = exports.LlamaCppClientAdapter = exports.fromEnvironment = exports.DEFAULT_RETRY_POLICY = exports.withRetry = exports.LLMService = void 0;
 // --- LLM Service ---
 var LLMService_1 = require("./llm/LLMService");
 Object.defineProperty(exports, "LLMService", { enumerable: true, get: function () { return LLMService_1.LLMService; } });
+// --- Retry Utilities ---
+var withRetry_1 = require("./shared/services/withRetry");
+Object.defineProperty(exports, "withRetry", { enumerable: true, get: function () { return withRetry_1.withRetry; } });
+Object.defineProperty(exports, "DEFAULT_RETRY_POLICY", { enumerable: true, get: function () { return withRetry_1.DEFAULT_RETRY_POLICY; } });
 // Export all core request/response/config types from the LLM module
 __exportStar(require("./llm/types"), exports);
 // Export all client adapter types
@@ -51,6 +55,7 @@ var parser_1 = require("./prompting/parser");
 Object.defineProperty(exports, "parseStructuredContent", { enumerable: true, get: function () { return parser_1.parseStructuredContent; } });
 Object.defineProperty(exports, "parseRoleTags", { enumerable: true, get: function () { return parser_1.parseRoleTags; } });
 Object.defineProperty(exports, "extractInitialTaggedContent", { enumerable: true, get: function () { return parser_1.extractInitialTaggedContent; } });
+Object.defineProperty(exports, "extractMarkerDelimitedContent", { enumerable: true, get: function () { return parser_1.extractMarkerDelimitedContent; } });
 Object.defineProperty(exports, "parseTemplateWithMetadata", { enumerable: true, get: function () { return parser_1.parseTemplateWithMetadata; } });
 var config_1 = require("./llm/config");
 Object.defineProperty(exports, "createFallbackModelInfo", { enumerable: true, get: function () { return config_1.createFallbackModelInfo; } });

package/dist/llm/LLMService.d.ts

@@ -2,6 +2,7 @@ import type { ApiKeyProvider, PresetMode } from '../types';
 import type { Logger, LogLevel } from '../logging/types';
 import type { LLMChatRequest, LLMChatRequestWithPreset, LLMResponse, LLMFailureResponse, ProviderInfo, ModelInfo, ApiProviderId, LLMSettings, ModelContext, LLMMessage } from "./types";
 import type { ModelPreset } from "../types/presets";
+import { type RetryPolicy } from "../shared/services/withRetry";
 export type { PresetMode };
 /**
  * Options for configuring the LLMService
@@ -15,6 +16,32 @@ export interface LLMServiceOptions {
     logLevel?: LogLevel;
     /** Custom logger implementation. If provided, logLevel is ignored. */
     logger?: Logger;
+    /**
+     * Retry policy for transient failures (rate limits, 5xx, network errors, timeouts).
+     * Defaults: maxRetries 2, initialDelayMs 500, maxDelayMs 10000, backoffFactor 2,
+     * retryOnTimeout true. Provider SDK-internal retries are disabled — this layer
+     * is the single owner of retry behavior. Set `maxRetries: 0` to disable.
+     */
+    retry?: Partial<RetryPolicy> & {
+        /** Whether REQUEST_TIMEOUT failures are retryable (default true) */
+        retryOnTimeout?: boolean;
+    };
+    /** Default per-request timeout in ms (overridable per call). SDK defaults apply when unset. */
+    timeoutMs?: number;
+}
+/**
+ * Per-call options for LLMService.sendMessage
+ */
+export interface SendMessageOptions {
+    /**
+     * Abort signal to cancel the request (client-side — the provider may still
+     * process and bill an already-dispatched request). Aborts are never retried.
+     */
+    signal?: AbortSignal;
+    /** Per-request timeout in ms (overrides the service-level timeoutMs) */
+    timeoutMs?: number;
+    /** Per-request retry cap (overrides the service-level retry.maxRetries) */
+    maxRetries?: number;
 }
 /**
  * Result from createMessages method
@@ -46,6 +73,8 @@ export declare class LLMService {
     private requestValidator;
     private settingsManager;
     private modelResolver;
+    private retryOptions;
+    private defaultTimeoutMs?;
     constructor(getApiKey: ApiKeyProvider, options?: LLMServiceOptions);
     /**
      * Gets list of supported LLM providers
@@ -66,7 +95,7 @@ export declare class LLMService {
      * @param request - The LLM chat request
      * @returns Promise resolving to either success or failure response
      */
-    sendMessage(request: LLMChatRequest | LLMChatRequestWithPreset): Promise<LLMResponse | LLMFailureResponse>;
+    sendMessage(request: LLMChatRequest | LLMChatRequestWithPreset, callOptions?: SendMessageOptions): Promise<LLMResponse | LLMFailureResponse>;
     /**
      * Gets all configured model presets
      *