datasaudi-mcp is an independent open-source project that serves as a client for DataSaudi, enabling Claude, Cursor, and any AI client that supports MCP (Model Context Protocol) to directly query Saudi Arabia's economic and social statistics in English or Arabic. The project's purpose is "refuse silent misdirection," meaning that data tools should clearly communicate the boundaries of their capabilities and, when out of scope, state so directly rather than guess.
[Purpose] This project is used to obtain Saudi official economic and social statistical indicators from within an AI client. Users do not need to know dataset names or query syntax; they only need to ask questions in natural language, and the model will automatically discover suitable datasets, retrieve the data, and answer. The data source is DataSaudi's Tesseract API (operated by Saudi Arabia's Ministry of Economy and Planning, with data produced by GASTAT), and each query is made in real time, so results stay in sync with DataSaudi itself.
[Main Capabilities]
- Covers 277 curated GASTAT / DataSaudi datasets, all at the headline indicator level.
- Economic: GDP (by industry/expenditure/activity), CPI and inflation (national and by province), producer price index, foreign direct investment, foreign trade (by country/product/flow), trade balance, and business and consumer confidence.
- Social: population (by province/gender/nationality/age), households and housing, wages, employment and participation rates, education, health, disability.
- Real estate: price indices by province and property type (villa/apartment/land), building permits, construction cost index, housing tenure and type.
- Other: tourism and hotel occupancy rates, energy and water resources, Hajj and Umrah, SAMA financial series, international reporting indicators (World Bank/United Nations).
- Arabic is a first-class feature: you can ask questions in Arabic and receive answers in Arabic, with member names returned in Arabic (such as الرياض for Riyadh, مكة المكرمة for Mecca, المنطقة الشرقية for the Eastern Province); catalog search also supports matching Arabic terms. Numbers and structure are consistent across languages; only human-readable titles change.
[How to Use]
- Prerequisites: uv must be installed. No API key, no database, no configuration required.
- In Claude Code, it can be configured with a single command, without editing files.
- In Claude Desktop: go to Settings → Developer → Edit Config, add the configuration block, then fully quit and reopen Claude Desktop (you must quit completely from the tray; just closing the window is not enough).
- In Cursor: go to Settings → Tools & MCP → New MCP Server, or edit ~/.cursor/mcp.json, and add the same configuration block.
- Once configured, you can ask questions directly, for example: "What was the real-estate price index in Riyadh by quarter, and how does it compare to the national trend?" or ask in Arabic "اعرض الرقم القياسي لأسعار العقارات في الرياض حسب الربع".
- The three tools do the actual work and are called automatically by the model.
[Use cases] Suitable for analysts, researchers, journalists, or developers who need quick access to Saudi macroeconomic and social statistics data, completing data queries and preliminary comparisons in natural language within an AI client, for example comparing inflation across provinces, analyzing real-estate price trends, viewing population structure, or trade partner rankings.
[Limitations]
- Only provides the real-estate price index (base year 100), not per-transaction prices in riyals (the latter belongs to MoJ / Srem).
- Does not include MoJ transaction volumes, REGA, municipal / ZATCA / CMA data, and does not provide microdata; this is a headline indicator layer of about 277 series, not a complete national corpus.
- The geographic granularity of most series stops at the provincial level, with no city/district breakdown.
- For out-of-scope questions, the server will state so explicitly rather than guess.
[Technical notes] Data comes in real time from api.datasaudi.sa (DataSaudi Tesseract API); the server does not keep local copies and fetches in real time for each query; it only caches a lightweight dataset catalog in memory for use during the session. The project uses the MIT license, and contributions are welcome (see CONTRIBUTING.md). DataSaudi's data usage terms are set by its operator; this project is an independent open-source client and credits GASTAT / MEP as the data source.
[Troubleshooting tips] If it does not work, check: whether you fully quit and reopened the application (just closing the window is not enough);whether the configuration is valid JSON (extra commas will break the file; use a JSON validator to check); whether uv is installed and in PATH (run uvx --version in the terminal); check the error message next to datasaudi in the client's MCP / tools panel.
