### Ollama Setup: Install and Pull Model

Source: https://context7.com/eslazarev/ai-wildberries-review-responder/llms.txt

Installs Ollama and downloads a specified model (e.g., gemma3:4b) for local LLM inference.

```bash
# Установка Ollama и загрузка модели
curl -fsSL https://ollama.com/install.sh | sh
ollama pull gemma3:4b
```

--------------------------------

### Install and Initialize Yandex CLI

Source: https://github.com/eslazarev/ai-wildberries-review-responder/blob/main/README.md

Installs the Yandex Cloud CLI and initializes it for use. Follow the official documentation for detailed setup.

```bash
curl https://storage.yandexcloud.net/yandexcloud-yc/install.sh | bash
yc init
```

--------------------------------

### Ollama Setup: Environment Variables

Source: https://context7.com/eslazarev/ai-wildberries-review-responder/llms.txt

Configures environment variables to point the application to a local Ollama instance, including the base URL, model name, and API key.

```bash
# Настройка через переменные окружения
export LLM__BASE_URL='http://localhost:11434/v1'
export LLM__MODEL='gemma3:4b'
export LLM__API_KEY='ollama'
export WILDBERRIES__API_TOKEN='your_wb_token'
```

--------------------------------

### Application Settings (settings.yaml)

Source: https://github.com/eslazarev/ai-wildberries-review-responder/blob/main/README.md

Example structure for the main application configuration file, settings.yaml. Defines settings for Wildberries API and LLM provider.

```yaml
wildberries:
  base_url: "https://feedbacks-api.wildberries.ru"
  request_timeout: 10
  batch_size: 10
  check_every_minutes: 30

llm:
  model: "gpt://{FOLDER_ID}/aliceai-llm/latest"
  base_url: "https://rest-assistant.api.cloud.yandex.net/v1"
  temperature: 0.3
  max_tokens: 600
  instructions: "..."
  prompt_template: "..."
  timeout: 10
```

--------------------------------

### Install Dependencies

Source: https://github.com/eslazarev/ai-wildberries-review-responder/blob/main/README.md

Installs project dependencies using pip within a virtual environment. Ensure you have Python and venv set up.

```bash
python -m venv .venv
source .venv/bin/activate
pip install -r requirements-dev.txt
```

--------------------------------

### Deploy AI Responder with npm

Source: https://github.com/eslazarev/ai-wildberries-review-responder/blob/main/README.md

Installs dependencies and deploys the function to Yandex Cloud Functions. Requires WILDBERRIES__API_TOKEN to be set.

```bash
npm install
WILDBERRIES__API_TOKEN='your_wb_token' serverless deploy
```

--------------------------------

### Local Ollama Configuration (Environment Variables)

Source: https://github.com/eslazarev/ai-wildberries-review-responder/blob/main/README.md

Example configuration for using a local Ollama model via environment variables. Sets the base URL, model name, and API key.

```bash
export LLM__BASE_URL='http://localhost:11434/v1'
export LLM__MODEL='gemma3:4b'
export LLM__API_KEY='ollama'
```

--------------------------------

### Yandex Cloud Deployment: Install Dependencies

Source: https://context7.com/eslazarev/ai-wildberries-review-responder/llms.txt

Installs the necessary Node.js dependencies for deploying with the Serverless Framework.

```bash
# Установка зависимостей
npm install
```

--------------------------------

### Run with Local Cron

Source: https://github.com/eslazarev/ai-wildberries-review-responder/blob/main/README.md

Starts the review processing with an internal cron scheduler. The interval is determined by `wildberries.check_every_minutes`. Requires API tokens.

```bash
export WILDBERRIES__API_TOKEN='your_wb_token'
export LLM__API_KEY='your_llm_api_key'
python -m src.entrypoints.docker_cron
```

--------------------------------

### Configure and Use OpenAIClient for LLM Interactions

Source: https://context7.com/eslazarev/ai-wildberries-review-responder/llms.txt

Configure OpenAIClient with settings for YandexGPT, OpenAI, or Ollama. Instantiate the client with specific parameters like temperature, max tokens, system instructions, and timeout. Use generate_reply to get responses.

```python
from src.infra.llm.openai_client import OpenAIClient
from src.infra.config.settings import LLMRuntimeSettings

# Конфигурация для YandexGPT
yandex_config = LLMRuntimeSettings(
    api_key="your-iam-token-or-api-key",
    model="gpt://folder-id/aliceai-llm/latest",
    base_url="https://rest-assistant.api.cloud.yandex.net/v1"
)

# Конфигурация для OpenAI
openai_config = LLMRuntimeSettings(
    api_key="sk-your-openai-key",
    model="gpt-4o-mini",
    base_url="https://api.openai.com/v1"
)

# Конфигурация для локальной Ollama
ollama_config = LLMRuntimeSettings(
    api_key="ollama",  # Любое непустое значение
    model="gemma3:4b",
    base_url="http://localhost:11434/v1"
)

# Создание клиента
llm_client = OpenAIClient(
    config=openai_config,
    temperature=0.3,
    max_tokens=600,
    instructions="Ты — вежливый сотрудник поддержки реселлера на Wildberries.",
    timeout=10
)

# Генерация ответа на промпт
prompt = """Ответь на отзыв Wildberries в дружелюбном тоне.
Full review payload:
{"text": "Сумка отличная!", "productValuation": 5, "has_photos": true}"""

reply = llm_client.generate_reply(prompt)
print(reply)
# "Благодарим за отзыв и приложенные фотографии! Рады, что сумка вам понравилась."
```

--------------------------------

### Local Ollama Configuration (YAML)

Source: https://github.com/eslazarev/ai-wildberries-review-responder/blob/main/README.md

Example configuration for using a local Ollama model via settings.yaml. Sets the base URL, model name, and API key.

```yaml
llm:
  base_url: "http://localhost:11434/v1"
  model: "gemma3:4b"
  api_key: "ollama"
```

--------------------------------

### Docker Entrypoint: Cron Schedule

Source: https://context7.com/eslazarev/ai-wildberries-review-responder/llms.txt

Starts a blocking process that uses APScheduler to run review processing at a defined interval. Suitable for continuous operation within a container.

```python
# Запуск с cron-расписанием (docker_cron)
from src.entrypoints.docker_cron import main

main()
```

--------------------------------

### Initialize and Use WildberriesClient

Source: https://context7.com/eslazarev/ai-wildberries-review-responder/llms.txt

Shows how to initialize the WildberriesClient with settings and use its methods to fetch reviews and publish replies. Requires the settings object to be configured.

```python
from src.infra.clients.wildberries import WildberriesClient
from src.infra.config.settings import Settings

# Инициализация клиента с настройками
settings = Settings()
wb_client = WildberriesClient(settings)

# Получение списка неотвеченных отзывов
reviews = wb_client.fetch_reviews()
# Возвращает: List[Review] — список доменных объектов Review

for review in reviews:
    print(f"ID: {review.id}")
    print(f"Текст: {review.text}")      # JSON-сериализованный payload отзыва
    print(f"Сводка: {review.summary}")  # Объединение text, pros, cons

# Публикация ответа на конкретный отзыв
wb_client.publish_reply(
    review_id="wb-123456",
    message="Благодарим за отзыв! Рады, что товар вам понравился."
)
```

--------------------------------

### Initialize Application Components for Review Response Orchestration

Source: https://context7.com/eslazarev/ai-wildberries-review-responder/llms.txt

Initialize all necessary components for the main application logic, including settings, logger, Wildberries client, LLM client, prompt builder, and reply generator. This sets up the core pipeline for processing reviews.

```python
from src.application.respond_on_reviews import respond_on_reviews
from src.infra.clients.wildberries import WildberriesClient
from src.infra.reply_generator import LLMReplyGenerator
from src.infra.prompt import PromptBuilder
from src.infra.llm.base import make_llm_client
from src.infra.config.settings import Settings
from src.infra.logger import init_logger

# Инициализация всех компонентов
settings = Settings()
logger = init_logger()

wildberries_client = WildberriesClient(settings)
llm_client = make_llm_client(settings)
prompt_builder = PromptBuilder(template=settings.llm.prompt_template)
reply_generator = LLMReplyGenerator(prompt_builder, llm_client)

```

--------------------------------

### Docker Entrypoint: Run Once

Source: https://context7.com/eslazarev/ai-wildberries-review-responder/llms.txt

Executes a single iteration of review processing and then exits. Suitable for one-off tasks.

```python
# Разовый запуск (docker_once)
from src.entrypoints.docker_once import run_once

run_once()
```

--------------------------------

### Deploy with Optional Parameters

Source: https://github.com/eslazarev/ai-wildberries-review-responder/blob/main/README.md

Deploys the application with optional environment variables for LLM API key and check interval. This allows customization during deployment.

```bash
LLM__API_KEY='your_llm_api_key' WILDBERRIES__API_TOKEN='your_wb_token' serverless deploy
WILDBERRIES__CHECK_EVERY_MINUTES='15' WILDBERRIES__API_TOKEN='your_wb_token' serverless deploy
```

--------------------------------

### Load Application Settings with Settings Class

Source: https://context7.com/eslazarev/ai-wildberries-review-responder/llms.txt

Demonstrates how to load application settings using the Settings class, which supports environment variables, .env files, file secrets, and settings.yaml. Accesses Wildberries API and LLM provider configurations.

```python
from src.infra.config.settings import Settings

# Загрузка настроек с автоматическим поиском settings.yaml и .env
settings = Settings()

# Доступ к настройкам Wildberries API
print(settings.wildberries.api_token)      # Токен API Wildberries
print(settings.wildberries.base_url)        # https://feedbacks-api.wildberries.ru
print(settings.wildberries.batch_size)      # Количество отзывов за итерацию (по умолчанию 10)
print(settings.wildberries.request_timeout) # Таймаут HTTP-запросов (по умолчанию 10 сек)
print(settings.wildberries.check_every_minutes) # Интервал cron (по умолчанию 30 мин)

# Доступ к настройкам LLM-провайдера
print(settings.llm.api_key)                 # API-ключ LLM (опционально для YC Functions)
print(settings.llm.model)                   # Идентификатор модели
print(settings.llm.base_url)                # URL API LLM-провайдера
print(settings.llm.temperature)             # Креативность ответа (по умолчанию 0.3)
print(settings.llm.max_tokens)              # Лимит токенов ответа (по умолчанию 600)
print(settings.llm.instructions)            # Системные инструкции для модели
print(settings.llm.prompt_template)         # Шаблон промпта для отзыва
```

--------------------------------

### Deploy with Custom LLM API Key

Source: https://github.com/eslazarev/ai-wildberries-review-responder/blob/main/README.md

Deploys the function while specifying a custom LLM API key. Requires WILDBERRIES__API_TOKEN.

```bash
LLM__API_KEY='your_llm_api_key' WILDBERRIES__API_TOKEN='your_wb_token' serverless deploy
```

--------------------------------

### Run Linting, Testing, and CI Checks

Source: https://github.com/eslazarev/ai-wildberries-review-responder/blob/main/README.md

Executes static analysis, security checks, and unit tests using make commands. These are typically part of the continuous integration pipeline.

```bash
make lint
make test
make ci
```

--------------------------------

### Yandex Cloud Deployment: Basic Deploy

Source: https://context7.com/eslazarev/ai-wildberries-review-responder/llms.txt

Deploys the application to Yandex Cloud using Serverless Framework. Requires the Wildberries API token to be set as an environment variable.

```bash
# Деплой с обязательным токеном Wildberries
WILDBERRIES__API_TOKEN='your_wb_token' serverless deploy
```

--------------------------------

### Create and Process WildberriesReview Object

Source: https://context7.com/eslazarev/ai-wildberries-review-responder/llms.txt

Illustrates the creation of a WildberriesReview object from raw data, accessing its computed properties, serializing it to a JSON payload, and converting it to a domain review entity.

```python
from src.infra.clients.wildberries_dto import WildberriesReview, ProductDetails

# Создание объекта отзыва из данных API
review_data = {
    "id": "fb-001",
    "productId": "12345",
    "userName": "Иван",
    "productValuation": 5,
    "text": "Отличный товар, рекомендую!",
    "pros": "Быстрая доставка",
    "cons": "Нет инструкции",
    "createdDate": "2024-01-15T10:30:00Z",
    "language": "ru",
    "productDetails": {
        "productName": "Сумка женская",
        "brandName": "BrandX",
        "supplierName": "SupplierY"
    },
    "photoLinks": [{"fullSize": "https://example.com/photo.jpg"}],
    "matchingSize": "Соответствует"
}

wb_review = WildberriesReview(**review_data)

# Вычисляемые свойства
print(wb_review.has_photos)     # True (есть фотографии)
print(wb_review.has_video)      # False (нет видео)
print(wb_review.summary)        # "Отличный товар...\nБыстрая доставка\nНет инструкции"

# Сериализация в JSON для промпта LLM
json_payload = wb_review.to_source_payload()
# {"id": "fb-001", "text": "...", "pros": "...", "has_photos": true, ...}

# Преобразование в доменную сущность
domain_review = wb_review.to_review()
# Review(id="fb-001", text=<json_payload>, summary=<summary>)
```

--------------------------------

### Build Contextualized Prompts with PromptBuilder

Source: https://context7.com/eslazarev/ai-wildberries-review-responder/llms.txt

Use PromptBuilder to create prompts from review data. It incorporates a custom template and the full review JSON payload for richer context, allowing LLMs to consider all review details.

```python
from src.infra.prompt import PromptBuilder
from src.domain.entities import Review

# Создание билдера с кастомным шаблоном
builder = PromptBuilder(
    template="""Отвечай на отзыв Wildberries на том же языке, что и отзыв.
Тон: дружелюбный, лаконичный, полезный. До 600 символов.
Не используй эмодзи."""
)

# Создание доменного отзыва
review = Review(
    id="review-123",
    text='{"text": "Качество хорошее", "productValuation": 4, "pros": "Цена"}',
    summary="Качество хорошее\nЦена"
)

# Построение финального промпта
prompt = builder.build(review)
print(prompt)
# Отвечай на отзыв Wildberries на том же языке, что и отзыв.
# Тон: дружелюбный, лаконичный, полезный. До 600 символов.
# Не используй эмодзи.
# Full review payload:
# {
#   "id": "review-123",
#   "text": "{\"text\": \"Качество хорошее\"...}",
#   "summary": "Качество хорошее\nЦена"
# }
```

--------------------------------

### Pull Docker Image

Source: https://github.com/eslazarev/ai-wildberries-review-responder/blob/main/README.md

Pulls the latest Docker image for the AI Wildberries Review Responder from Docker Hub.

```bash
docker pull eslazarev/ai-wildberries-review-responder:latest
```

--------------------------------

### Deploy with Custom Check Schedule

Source: https://github.com/eslazarev/ai-wildberries-review-responder/blob/main/README.md

Deploys the function and customizes the schedule for checking reviews. Requires WILDBERRIES__API_TOKEN.

```bash
WILDBERRIES__CHECK_EVERY_MINUTES='15' WILDBERRIES__API_TOKEN='your_wb_token' serverless deploy
```

--------------------------------

### Initiate Review Processing

Source: https://context7.com/eslazarev/ai-wildberries-review-responder/llms.txt

This function orchestrates the review processing pipeline, fetching reviews, generating replies, and publishing them.

```python
respond_on_reviews(
    review_fetcher=wildberries_client,
    reply_generator=reply_generator,
    review_publisher=wildberries_client,
    logger=logger
)
```

--------------------------------

### Run Single Processing Iteration

Source: https://github.com/eslazarev/ai-wildberries-review-responder/blob/main/README.md

Executes a single run of the review processing logic. Requires Wildberries and LLM API tokens to be set as environment variables.

```bash
export WILDBERRIES__API_TOKEN='your_wb_token'
export LLM__API_KEY='your_llm_api_key'
python -m src.entrypoints.docker_once
```

--------------------------------

### Configuration Settings

Source: https://context7.com/eslazarev/ai-wildberries-review-responder/llms.txt

Defines default settings for Wildberries API interaction and LLM parameters. Environment variables can override these values.

```yaml
# settings.yaml
wildberries:
  base_url: "https://feedbacks-api.wildberries.ru"
  request_timeout: 10
  batch_size: 10
  check_every_minutes: 30

llm:
  model: "gpt://{FOLDER_ID}/aliceai-llm/latest"
  base_url: "https://rest-assistant.api.cloud.yandex.net/v1"
  temperature: 0.3
  max_tokens: 600
  instructions: |
    Ты — вежливый сотрудник поддержки реселлера, отвечающий на отзывы Wildberries.
  prompt_template: |
    Отвечай на отзыв Wildberries на том же языке, что и отзыв.
    Тон: дружелюбный, лаконичный, полезный. До 600 символов.
    Ты не работаешь в Wildberries, не упоминай компанию в ответе.
    Не используй эмодзи.
  timeout: 10
```

--------------------------------

### Configure LLM Provider

Source: https://github.com/eslazarev/ai-wildberries-review-responder/blob/main/README.md

Sets environment variables to connect to an OpenAI-compatible LLM provider. This allows using different LLM services by changing the base URL, model, and API key.

```bash
export LLM__BASE_URL='https://api.openai.com/v1'
export LLM__MODEL='gpt-4o-mini'
export LLM__API_KEY='your_key'
```

--------------------------------

### Generate Replies using LLMReplyGenerator

Source: https://context7.com/eslazarev/ai-wildberries-review-responder/llms.txt

LLMReplyGenerator combines PromptBuilder and LLMClient to generate review replies. It takes a domain Review object and returns the final reply text, integrating prompt construction and LLM interaction.

```python
from src.infra.reply_generator import LLMReplyGenerator
from src.infra.prompt import PromptBuilder
from src.infra.llm.base import make_llm_client
from src.infra.config.settings import Settings
from src.domain.entities import Review

settings = Settings()

# Создание компонентов
prompt_builder = PromptBuilder(template=settings.llm.prompt_template)
llm_client = make_llm_client(settings)  # Фабрика создает OpenAIClient

# Сборка генератора ответов
reply_generator = LLMReplyGenerator(
    prompt_builder=prompt_builder,
    llm_client=llm_client
)

# Генерация ответа на отзыв
review = Review(
    id="wb-456",
    text='{"text": "Товар пришел с дефектом", "productValuation": 2}',
    summary="Товар пришел с дефектом"
)

reply = reply_generator.generate(review)
print(reply)
# "Приносим извинения за неудобства. Пожалуйста, свяжитесь с нами для замены товара."
```

--------------------------------

### Run Docker Container for Single Iteration

Source: https://github.com/eslazarev/ai-wildberries-review-responder/blob/main/README.md

Runs the Docker container to perform a single review processing task. Specify `src.entrypoints.docker_once` to override the default scheduled mode. Requires API tokens.

```bash
docker run --rm \
  -e WILDBERRIES__API_TOKEN='your_wb_token' \
  -e LLM__API_KEY='your_llm_api_key' \
  eslazarev/ai-wildberries-review-responder:latest src.entrypoints.docker_once
```

--------------------------------

### Run Docker Container with Internal Schedule

Source: https://github.com/eslazarev/ai-wildberries-review-responder/blob/main/README.md

Runs the Docker container in its default mode, which includes an internal scheduler. Configure the check interval using environment variables. Requires API tokens.

```bash
docker run --rm \
  -e WILDBERRIES__API_TOKEN='your_wb_token' \
  -e WILDBERRIES__CHECK_EVERY_MINUTES='15' \
  -e LLM__API_KEY='your_llm_api_key' \
  eslazarev/ai-wildberries-review-responder:latest
```

--------------------------------

### Ollama Local Run

Source: https://context7.com/eslazarev/ai-wildberries-review-responder/llms.txt

Executes the review processing using a locally running Ollama model. This command initiates a single run.

```python
# Локальный запуск
python -m src.entrypoints.docker_once
```

--------------------------------

### Docker Run Command: One-Time Execution

Source: https://context7.com/eslazarev/ai-wildberries-review-responder/llms.txt

Command to run a Docker container for a single review processing task. It mounts environment variables for authentication and API keys.

```bash
# docker run --rm \
#   -e WILDBERRIES__API_TOKEN='your_wb_token' \
#   -e LLM__API_KEY='your_llm_api_key' \
#   eslazarev/ai-wildberries-review-responder:latest \
#   src.entrypoints.docker_once
```

--------------------------------

### Yandex Cloud Deployment: Custom LLM Key

Source: https://context7.com/eslazarev/ai-wildberries-review-responder/llms.txt

Deploys the application with a custom LLM API key, such as for OpenAI. This overrides the default YandexGPT configuration.

```bash
# Деплой с кастомным LLM-ключом (для OpenAI)
LLM__API_KEY='sk-openai-key' \
WILDBERRIES__API_TOKEN='your_wb_token' \
serverless deploy
```

--------------------------------

### Run Architectural Tests

Source: https://github.com/eslazarev/ai-wildberries-review-responder/blob/main/README.md

Executes architectural tests using pytest. The `-m architecture` flag filters tests specifically for architectural validation.

```bash
pytest -m architecture
```

--------------------------------

### Yandex Cloud Deployment: Change Check Interval

Source: https://context7.com/eslazarev/ai-wildberries-review-responder/llms.txt

Deploys the application with a modified interval for checking reviews. The WILDBERRIES__CHECK_EVERY_MINUTES variable controls this setting.

```bash
# Изменение интервала проверки на 15 минут
WILDBERRIES__CHECK_EVERY_MINUTES='15' \
WILDBERRIES__API_TOKEN='your_wb_token' \
serverless deploy
```

--------------------------------

### Yandex Cloud Functions Entrypoint

Source: https://context7.com/eslazarev/ai-wildberries-review-responder/llms.txt

The handler function for Yandex Cloud Functions. It automatically uses FOLDER_ID for YandexGPT and an IAM token from the context if LLM__API_KEY is not set. Returns a JSON status.

```python
# src/entrypoints/yandex_cloud_function.py
from src.entrypoints.yandex_cloud_function import handler

# Вызов из Yandex Cloud Functions (event и context передаются платформой)
event = {}
context = type('Context', (), {
    'function_folder_id': 'b1g123456789',
    'token': {'access_token': 'iam-token-from-yc'}
})()

result = handler(event, context)
print(result)
```

--------------------------------

### Yandex Cloud Deployment: Remove Resources

Source: https://context7.com/eslazarev/ai-wildberries-review-responder/llms.txt

Removes all deployed resources associated with the Serverless Framework application from Yandex Cloud.

```bash
# Удаление всех ресурсов
serverless remove
```

--------------------------------

### Docker Run Command: Cron Schedule

Source: https://context7.com/eslazarev/ai-wildberries-review-responder/llms.txt

Command to run a Docker container in detached mode with an internal cron schedule. The WILDBERRIES__CHECK_EVERY_MINUTES variable sets the check interval.

```bash
# docker run -d \
#   -e WILDBERRIES__API_TOKEN='your_wb_token' \
#   -e WILDBERRIES__CHECK_EVERY_MINUTES='15' \
#   -e LLM__API_KEY='your_llm_api_key' \
#   eslazarev/ai-wildberries-review-responder:latest
```

--------------------------------

### Remove Yandex Cloud Resources

Source: https://github.com/eslazarev/ai-wildberries-review-responder/blob/main/README.md

Removes all deployed resources associated with the serverless application from Yandex Cloud.

```bash
serverless remove
```

=== COMPLETE CONTENT === This response contains all available snippets from this library. No additional content exists. Do not make further requests.