mcp-leclerc-drive
MCP server for E.Leclerc Drive that enables searching products, managing a cart, and preparing grocery orders natively through natural language.
README
mcp-leclerc-drive
The first open-source MCP server for E.Leclerc Drive β let Claude search products, manage a cart, and prepare grocery orders natively, instead of clicking through the website.
π’ v0.1 β working. All five tools are implemented and validated end-to-end against the live site (store 053701): search, add, read, update, remove. Auth reads your existing Chrome session automatically (no copy-paste). See
docs/api-capture.mdfor the reverse-engineered API.
Why
E.Leclerc Drive has no public API. Today the only way to automate it is browser automation β slow (~3β5 s per item) and fragile (blind clicks). This project exposes the underlying operations as proper MCP tools so any MCP client (Claude Desktop, Claude Code) can drive it directly.
Tools
| Tool | Description |
|---|---|
search_product(query) |
Search the catalogue β products with price, price/kg, Nutri-Score, availability, and an id. |
add_to_cart(product_id, quantity?) |
Add a product to the cart. |
remove_from_cart(product_id) |
Remove a line from the cart. |
update_quantity(product_id, quantity) |
Set a line's quantity (0 removes it). |
get_cart() |
Read the full cart with total. |
Status
- [x] MCP server scaffold (stdio,
@modelcontextprotocol/sdk) - [x] Tool contracts (
search_product,add_to_cart,remove_from_cart,update_quantity,get_cart) - [x] Cookie-based auth model
- [x] Reverse-engineer Leclerc Drive endpoints (validated live β see
docs/api-capture.md) - [x] Wire endpoints into
src/leclerc/client.ts - [x] Auto-read auth cookie from the local Chrome session (
src/auth/cookies.ts) - [x] End-to-end validation of all five tools against the live store β
- [ ] Test under Claude Desktop / Claude Code (MCP client integration)
- [ ] Handle DataDome cookie refresh / session expiry gracefully
- [ ] Publish to npm + submit to MCP registry
Install (development)
git clone https://github.com/skunkobi/mcp-leclerc-drive.git
cd mcp-leclerc-drive
npm install
npm run build
Configuration & auth
Default (recommended): borrow your Chrome session. Log into Leclerc Drive in
Chrome once. The server reads the session cookie (incl. the datadome cookie)
directly from your local Chrome profile β no copy-paste, and it refreshes itself
as your browser session does. On macOS the first read triggers a one-time
Keychain prompt ("Chrome Safe Storage"); approve it. The server must run on the
same machine as Chrome.
Headless deploys (VPS / CI): set LECLERC_COOKIE to a captured Cookie
header and it takes precedence over Chrome (note: a captured DataDome cookie
expires, so this needs periodic refreshing).
| Env var | Default | Description |
|---|---|---|
LECLERC_STORE_ID |
053701 |
Store id (La Ville-aux-Dames). |
LECLERC_HOST |
fd9-courses.leclercdrive.fr |
Backend host (the fdN prefix varies by store). |
LECLERC_CHROME_PROFILE |
Default |
Chrome profile directory to read cookies from. |
LECLERC_COOKIE |
β | Optional raw Cookie override; skips Chrome when set. |
LECLERC_MIN_INTERVAL_MS |
1000 |
Minimum delay between two requests (anti-strike). |
LECLERC_JITTER_MS |
400 |
Extra random jitter added between requests. |
LECLERC_MAX_RETRIES |
3 |
Retries on a 403/429 before giving up. |
LECLERC_BACKOFF_BASE_MS |
1500 |
Base retry backoff (doubles each attempt). |
Staying under DataDome (anti-strike)
Leclerc Drive is protected by DataDome, which blocks (HTTP 403) traffic that looks automated β especially bursts of parallel requests. The server defends against this automatically so you don't get struck:
- Serialized requests β every call goes through a single queue, one at a time, so even if several tools are invoked "in parallel" they never hit the site at once.
- Spacing + jitter β a ~1 s pause (plus random jitter) between requests.
- Retry with backoff β a 403/429 is retried a few times with exponential
backoff, re-reading a fresh cookie from Chrome each attempt (a real browser
refreshes its
datadomecookie on its own).
If you ever do get a persistent 403, just open Leclerc Drive in Chrome to
refresh your session and retry. Tune the cadence with the LECLERC_* env vars
above.
Finding your store id and host: open your Drive in a browser β the URL looks
like https://fd9-courses.leclercdrive.fr/magasin-053701-053701-Your-Town/. The
6-digit number is your LECLERC_STORE_ID; the fdN-courses.leclercdrive.fr part
is your LECLERC_HOST (the fdN prefix varies by region). The defaults point to
store 053701 (La Ville-aux-Dames).
Claude Desktop / Claude Code (mcp config)
{
"mcpServers": {
"leclerc-drive": {
"command": "node",
"args": ["/absolute/path/to/mcp-leclerc-drive/dist/index.js"],
"env": {
"LECLERC_STORE_ID": "053701"
}
}
}
}
(No cookie needed in the config β it comes from your Chrome session. Just be logged into Leclerc Drive in Chrome.)
Development
npm run dev # tsc --watch
npm run typecheck # type-check without emitting
npm run inspect # run under the MCP Inspector
Architecture
src/
index.ts # MCP server: registers the 5 tools over stdio
config.ts # env-based config (store, host, cookie source)
types.ts # Product / CartItem / Cart
auth/
cookies.ts # cookie provider: auto-read from Chrome, env override
leclerc/
client.ts # Leclerc Drive backend client (search + cart, validated)
throttle.ts # anti-strike: serialize + space out + retry (DataDome)
docs/
api-capture.md # the reverse-engineered Leclerc Drive API
Contributing
This is a community tool β contributions are very welcome, whether it's a bug fix, support for your store, or a whole new capability (checkout, delivery slots, saved listsβ¦).
See CONTRIBUTING.md for dev setup, how to smoke-test
against your own account (npm run smoke), and β most useful for this project β
a short guide on how to reverse-engineer a new Leclerc Drive endpoint and
wire it in. Good first issues are listed in the status checklist above.
Feedback & contact
Feedback, bug reports, and ideas are very welcome β this is an early v0.1.
- Issues / PRs: open an issue on the repo.
- Email: alexandreyagoubi@gmail.com
Disclaimer
Unofficial. Not affiliated with or endorsed by E.Leclerc. Use with your own account, at your own risk, in line with the site's terms of service. Intended for personal automation of your own grocery shopping.
License
MIT
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.