digikala

The Digikala MCP Server provides an intelligent product discovery workflow designed for Digikala's ecosystem. It exposes a bilingual search capability (Persian and English) and returns filtered, marketable product data with pricing in Tooman and Rial. Core tools include get_optimized_keywords_and_categories for keyword/category optimization, search_products for filtering and sorting results, get_product_details for deep-dives on promising items, get_product_recommendations for similar options, and search_text_lenz for AI-powered visual/semantic searches. Use get_optimized_keywords_and_categories first to obtain category IDs and optimized keywords, then pair those with search_products to retrieve high-quality results. When you identify promising items, you can drill down with get_product_details and expand with get_product_recommendations to refine your discovery path.

Prerequisites:

• Python 3.8+ environment
• uv (MCP CLI) installed or available in PATH
• Access to the repository containing the Digikala MCP Server

Installation steps:

1. Install MCP runtime/CLI if not already present:
   • If using uv, ensure it is installed and accessible in your shell.
2. Clone the repository containing the Digikala MCP Server:
   • git clone https://github.com/your org/rezashahnazar-digikala-mcp-server.git
3. Install Python dependencies (adjust if you use a virtual environment):
   • cd rezashahnazar-digikala-mcp-server
   • python -m pip install -r requirements.txt
4. Run the MCP server locally for testing:
   • uv run python main.py
5. Optional: Integrate with Claude Desktop or other clients using the provided configuration snippet in the README, adapted to your environment.

Notes:

• Ensure that your environment has network access to Digikala APIs or proxies used by the MCP server.
• Set environment variables for logging or API keys if required by the deployment environment.

Tips and caveats:

• All price inputs/outputs use Tooman; conversions to Rial are performed internally (1 Tooman = 10 Rials).
• Price filtering in search_products uses Tooman values; ensure you convert budgets accordingly.
• Ratings are shown only when there are at least 10 reviews; otherwise rating fields may be null.
• The discovery workflow favors bilingual input to maximize keyword/category coverage; call get_optimized_keywords_and_categories separately for Persian and English queries.
• When integrating with Claude Desktop, ensure the VP/paths in the config reflect your local setup and that main.py is reachable from the specified working directory.