Codex is unable to use playwright mcp with extension?

[MikeK184]

Hi there,

I was using Playwright-MCP with Claude Code/Gemini CLI for a while now, I recently switched to Codex from OpenAI, it seems like I can't use the playwright mcp to navigate to open sessions or have it start its own browser to navigate to any pages?

>_ You are using OpenAI Codex in ~/termedev

 To get started, describe a task or try one of these commands:

 /status - show current session configuration and token usage
 /approvals - choose what Codex can do without approval
 /model - choose what model and reasoning effort to use

/mcp

🔌  MCP Tools

  • Server: context7
    • Command: npx -y @upstash/context7-mcp
    • Tools: get-library-docs, resolve-library-id

  • Server: dbhub-mariadb
    • Command: npx -y @bytebase/dbhub --transport stdio --dsn mariadb://wp_user:wp_pass@127.0.0.1:3306/wp_dev
    • Tools: execute_sql

  • Server: playwright
    • Command: npx @playwright/mcp@latest --extension
    • Tools: browser_click, browser_close, browser_console_messages, browser_drag, browser_evaluate, browser_file_upload, browser_fill_form, browser_handle_dialog, browser_hover, browser_install, browser_navigate, browser_navigate_back, browser_network_requests, browser_press_key, browser_resize,
browser_select_option, browser_snapshot, browser_tabs, browser_take_screenshot, browser_type, browser_wait_for

  • Server: serena
    • Command: uvx --from git+https://github.com/oraios/serena serena start-mcp-server --context ide-assistant --project $(pwd)
    • Tools: activate_project, check_onboarding_performed, delete_memory, find_file, find_referencing_symbols, find_symbol, get_symbols_overview, insert_after_symbol, insert_before_symbol, list_dir, list_memories, onboarding, read_memory, replace_symbol_body, search_for_pattern,
think_about_collected_information, think_about_task_adherence, think_about_whether_you_are_done, write_memory


▌use playwright to navigate to google.com

> Opening a browser and navigating to google.com now.

tool running...
playwright.browser_navigate({"url":"https://www.google.com"})

tool failed, duration: 5.03s
playwright.browser_navigate({"url":"https://www.google.com"})

### ResultError: Extension connection timeout. Make sure the "Playwright MCP Bridge" extension is installed. See https://github.com/microsoft/playwright-mcp/blob/main/extension/README.md for installation instructions.

🖐  Tell the model what to do differently

However the extensions is installed on Chrome and Edge

As a side note, I am running codex in wsl2 with ubuntu 24, however this was working with CC/Gemini CLI.

Does anyone have any ideas how to proceed?

[pavelfeldman]

Not sure why it would not work. Pleases update this issue if you figure it out, closing it as non-actionable on Playwright MCP side.

[chrisbjr]

I can't get it to show a browser as well. In Claude Code and OpenCode, this works perfectly in my Windows 11 running in WSL2 ubuntu but i can't seem to force codex to open a headed browser and let it show me what it's doing

[pinkpanther303]

Same for me on Mac OS running it into VSC Terminal

[alooshxl]

Same for me it work sometimes and most of the time no

[psychananaz]

Only variables listed in env_args are passed to the server process, since codex uses an allow-list feature for environment variables.

Headful Chromium requires an X server to run. If DISPLAY is not set, Chromium won't start.

The configuration should look like this:

[mcp_servers.playwright]
command = "npx"
args = ["@playwright/mcp@latest"]
env_vars = ["DISPLAY"]

This resolved the issue for me on Linux (wayland).