Ozon seller price monitoring with auto-login
OZON MCP Server
A Playwright-based MCP server for the Ozon seller cabinet: automatic login via a QQ Mail code and a check for marketing-action prices below the minimum
Medium risk
We rate an entry medium when the tool runs code, makes network calls or reads project files. Check what exactly it does before installing.
Why this level
- Automatically logs into the Ozon seller cabinet and stores a mail app code in .env
- Keeps a local browser profile with an active account session
Install
Manual install
uv tool install ozon-mcp && claude mcp add ozon --transport stdio -- ozon-mcpInstalling the package from PyPI and registering it as an MCP server in Claude Code.
This is third-party code. Review the repository files before installing.
What it does
The server keeps a Playwright Chrome session and, via the login-with-email-code tool, signs itself into seller.ozon.ru by fetching a one-time code from a QQ Mail inbox over IMAP with a supplied app password. The get-marketing-actions tool reads product data from current marketing actions one page or all pages at a time, to find items priced below the allowed minimum. The same functionality is available outside the MCP protocol through a standalone cli.py, with CSV or JSON export and a headless mode. The Chrome profile is saved locally after the first login so verification doesn't repeat.
Who it is for. For Ozon sellers using QQ Mail who need a quick check on whether a product's marketing-action price has dropped below the allowed minimum.
Good fit when
- You have a QQ Mail account tied to Ozon with IMAP enabled
- You need to regularly check marketing-action prices in your cabinet
- You are fine logging in via an email code rather than a direct password
Not a fit when
- Your Ozon-linked mailbox is not on QQ: the login flow is built specifically for it
- You don't want an email app password sitting in .env on disk
- You need to change something in the cabinet rather than just view action prices: the server doesn't do that
Example request
Log into my Ozon cabinet and show marketing-action products priced below the minimumLimitations
The README is in Chinese and English. Login is hard-tied to QQ Mail and a 16-digit app code fetched over IMAP/SMTP, it won't work with other mail providers. The server stores a saved Chrome session profile locally, and the README explicitly warns not to commit .env or chrome-profile/ to git.
How to disable. Run claude mcp remove ozon and delete the local .env and chrome-profile/ with the saved session.
MCP
- Transport
- stdio
- Authentication
- API key
| Environment variables | |
|---|---|
| ozon_username required, secret | The QQ Mail login tied to the Ozon account |
| qq_imap_auth_code required, secret | The 16-digit QQ Mail app authorization code for IMAP |
Security check
- Automatically logs into the Ozon seller cabinet and stores a mail app code in .env
- Keeps a local browser profile with an active account session
README in short
The Chinese and English README describes installing via PyPI or from source, setting up .env with the login and a 16-digit QQ app code, connecting to Claude Code and Cursor, both MCP tools with call examples, the equivalent cli.py with pagination and output flags, and a changelog for versions 0.1.0 and 0.1.2.
FAQ
Can the server change a product's price?
No, per the README only login-with-email-code and get-marketing-actions are available, neither changes cabinet data.
Is QQ Mail mandatory?
Yes, auto-login is built around fetching a code from a QQ Mail inbox over IMAP, the README describes no alternative for other mail services.
Related
Official DataLens (Yandex) skills for Claude Code, Codex and OpenCode: SDK, HTML reports and RLS resolution
A Yandex Metrika MCP server generated from the API spec: 108 methods, 10 tools declared by default, transparent filters and response metadata
An open MCP server and agent skill set for SEO: keyword research, competitors, backlinks and site audits powered by DataForSEO
A Claude Code plugin for SEO audits: technical SEO, E-E-A-T, schema, local and AI search via parallel subagents