ANSES Ciqual MCP Server

An MCP (Model Context Protocol) server providing SQL access to the ANSES Ciqual French food composition database. Query nutritional data for over 3,000 foods with full-text search support.

Features

• 🍎 Comprehensive Database: Access nutritional data for 3,185+ French foods

• 🔍 SQL Interface: Query using standard SQL with full flexibility

• 🌍 Bilingual Support: French and English food names

• 🔤 Fuzzy Search: Built-in full-text search with typo tolerance

• 📊 60+ Nutrients: Detailed composition including vitamins, minerals, macros, and more

• 🔄 Auto-Updates: Automatically refreshes data yearly from ANSES (checks on startup)

• 🔒 Read-Only: Safe queries with no risk of data modification

• 💾 Lightweight: ~10MB SQLite database with efficient indexing

Installation

Via pip

pip install ciqual-mcp

Via uvx (recommended)

uvx ciqual-mcp

From source

git clone https://github.com/zzgael/ciqual-mcp.git
cd ciqual-mcp
pip install -e .

MCP Client Configuration

Claude Desktop

Add to your Claude Desktop configuration:

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%/Claude/claude_desktop_config.json
Linux: ~/.config/Claude/claude_desktop_config.json

{
  "mcpServers": {
    "ciqual": {
      "command": "uvx",
      "args": ["ciqual-mcp"]
    }
  }
}

Gemini CLI

Add to your Gemini CLI configuration file ~/.gemini/settings.json:

{
  "mcpServers": {
    "ciqual": {
      "command": "uvx",
      "args": ["ciqual-mcp"]
    }
  }
}

Codex CLI

Add to your Codex CLI configuration file ~/.codex/config.toml:

[mcp_servers.ciqual]
command = "uvx"
args = ["ciqual-mcp"]

Usage

As an MCP Server

The server implements the Model Context Protocol and exposes a single query function:

# Start the server standalone (for testing)
ciqual-mcp

Direct Python Usage

from ciqual_mcp.data_loader import initialize_database

# Initialize/update the database
initialize_database()

# Then use SQLite directly
import sqlite3
conn = sqlite3.connect("~/.ciqual/ciqual.db")
cursor = conn.execute("SELECT * FROM foods WHERE alim_nom_eng LIKE '%apple%'")

API Documentation

MCP Function: query

The server exposes a single MCP function for executing SQL queries on the Ciqual database.

Function Signature

async def query(sql: str) -> list[dict]

Parameters

• sql (string, required): The SQL query to execute on the database

  • Must be a SELECT or WITH query (read-only access)

  • Supports all standard SQLite SQL syntax

  • Can use JOIN, GROUP BY, ORDER BY, etc.

  • Supports full-text search via the foods_fts table

Returns

• list[dict]: Array of result rows, where each row is a dictionary with column names as keys

  • Empty list if no results match the query

  • Error dictionary with "error" key if query fails

Error Handling

The function returns an error dictionary in these cases:

• Database not initialized: {"error": "Database not initialized..."}

• Non-SELECT query attempted: {"error": "Only SELECT queries are allowed for safety."}

• SQL syntax error: {"error": "SQL error: [details]"}

• Table not found: {"error": "Table not found. Available tables: foods, nutrients, composition, foods_fts, food_groups"}

Example Usage in MCP Context

{
  "method": "query",
  "params": {
    "sql": "SELECT f.alim_nom_eng, n.const_nom_eng, c.teneur, n.unit FROM foods f JOIN composition c ON f.alim_code = c.alim_code JOIN nutrients n ON c.const_code = n.const_code WHERE f.alim_nom_eng LIKE '%apple%' AND n.const_code IN (328, 25000, 31000)"
  }
}

Response Example

[
  {
    "alim_nom_eng": "Apple, raw",
    "const_nom_eng": "Energy",
    "teneur": 52.0,
    "unit": "kcal/100g"
  },
  {
    "alim_nom_eng": "Apple, raw",
    "const_nom_eng": "Protein",
    "teneur": 0.3,
    "unit": "g/100g"
  }
]

Database Schema

Tables

foods - Food items

• alim_code (INTEGER, PK): Unique food identifier

• alim_nom_fr (TEXT): French name

• alim_nom_eng (TEXT): English name

• alim_grp_code (TEXT): Food group code

nutrients - Nutrient definitions

• const_code (INTEGER, PK): Unique nutrient identifier

• const_nom_fr (TEXT): French name

• const_nom_eng (TEXT): English name

• unit (TEXT): Measurement unit (g/100g, mg/100g, etc.)

composition - Nutritional values

• alim_code (INTEGER): Food identifier

• const_code (INTEGER): Nutrient identifier

• teneur (REAL): Value per 100g

• code_confiance (TEXT): Confidence level (A/B/C/D)

foods_fts - Full-text search

Virtual table for fuzzy matching with French/English names

Common Nutrient Codes

Example Queries

Basic Search

-- Find foods by name
SELECT * FROM foods WHERE alim_nom_eng LIKE '%orange%';

-- Fuzzy search (handles typos)
SELECT * FROM foods_fts WHERE foods_fts MATCH 'orang*';

Nutritional Queries

-- Get vitamin C content for oranges
SELECT f.alim_nom_eng, c.teneur as vitamin_c_mg
FROM foods f
JOIN composition c ON f.alim_code = c.alim_code
WHERE f.alim_nom_eng LIKE '%orange%' 
  AND c.const_code = 55400;

-- Find foods highest in protein
SELECT f.alim_nom_eng, c.teneur as protein_g
FROM foods f
JOIN composition c ON f.alim_code = c.alim_code
WHERE c.const_code = 25000
ORDER BY c.teneur DESC
LIMIT 10;

-- Compare macros for different foods
SELECT 
    f.alim_nom_eng as food,
    MAX(CASE WHEN c.const_code = 25000 THEN c.teneur END) as protein_g,
    MAX(CASE WHEN c.const_code = 31000 THEN c.teneur END) as carbs_g,
    MAX(CASE WHEN c.const_code = 40000 THEN c.teneur END) as fat_g,
    MAX(CASE WHEN c.const_code = 328 THEN c.teneur END) as calories_kcal
FROM foods f
JOIN composition c ON f.alim_code = c.alim_code
WHERE f.alim_nom_eng IN ('Apple, raw', 'Banana, raw', 'Orange, raw')
  AND c.const_code IN (25000, 31000, 40000, 328)
GROUP BY f.alim_code, f.alim_nom_eng;

Dietary Restrictions

-- Find low-sodium foods (<100mg/100g)
SELECT f.alim_nom_eng, c.teneur as sodium_mg
FROM foods f
JOIN composition c ON f.alim_code = c.alim_code
WHERE c.const_code = 10110 
  AND c.teneur < 100
ORDER BY c.teneur ASC;

-- High-fiber foods (>5g/100g)
SELECT f.alim_nom_eng, c.teneur as fiber_g
FROM foods f
JOIN composition c ON f.alim_code = c.alim_code
WHERE c.const_code = 34100 
  AND c.teneur > 5
ORDER BY c.teneur DESC;

Data Source

Data is sourced from the official ANSES Ciqual database:

• Website: https://ciqual.anses.fr/

• Data portal: https://www.data.gouv.fr/fr/datasets/table-de-composition-nutritionnelle-des-aliments-ciqual/

The database is automatically updated yearly when the server starts (data hasn't changed since 2020, so yearly updates are sufficient).

Requirements

• Python 3.9 or higher

• 50MB free disk space (for database)

• Internet connection (for initial data download)

License

MIT License - See LICENSE file for details

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

Development

Running Tests

# Install development dependencies
pip install -e .
pip install pytest pytest-asyncio

# Run unit tests
python -m pytest tests/test_server.py -v

# Run functional tests (requires database)
python -m pytest tests/test_functional.py -v

Troubleshooting

Database not initializing

• Check internet connection

• Ensure write permissions to ~/.ciqual/ directory

• Try manual initialization: python -m ciqual_mcp.data_loader

XML parsing errors

• The tool handles malformed XML automatically with recovery mode

• If issues persist, delete ~/.ciqual/ciqual.db and restart

Credits

Developed by Gael Debost as part of GPT Workbench, a multi-LLM interface for medical research developed by Ideagency.

Data provided by ANSES (Agence nationale de sécurité sanitaire de l'alimentation, de l'environnement et du travail).

Citation

If you use this tool in your research, please cite:

@software{ciqual_mcp,
  title = {ANSES Ciqual MCP Server},
  author = {Gael Debost},
  year = {2025},
  url = {https://github.com/zzgael/ciqual-mcp}
}