dejo

Codex

Give OpenAI Codex a browser that remembers websites. The first time Codex does a task on a site, it explores and Dejo learns the route. After that, the same task is one tool call.

You need Python 3.11 or newer and Codex.

1. Install Dejo

pip install dejo-mcp
playwright install chromium

2. Add it to Codex

Either run:

codex mcp add dejo -- dejo-mcp

Or add this to ~/.codex/config.toml yourself:

[mcp_servers.dejo]
command = "dejo-mcp"

The Codex CLI, the IDE extension and the ChatGPT desktop app share this file, so Dejo shows up in all three.

3. Check it is connected

codex mcp list

You should see dejo. Inside Codex, /mcp lists the servers that are running. If Dejo is missing or fails to start, run dejo doctor.

4. Ask for something

Start a new Codex session and ask in your own words:

Go to github.com/microsoft/playwright and tell me how many open issues it has.

The first time, Codex explores the site with dejo_act and dejo_read, then saves the route with dejo_save. Ask again in a new session and dejo_run answers in one call.

Timeouts

Codex gives a server 10 seconds to start and 60 seconds per tool call, unless you change it. Most Dejo calls finish well inside that. If a slow site needs longer, raise it for Dejo only:

[mcp_servers.dejo]
command = "dejo-mcp"
tool_timeout_sec = 180

Settings

Dejo's settings are environment variables. In config.toml they go in an env table:

[mcp_servers.dejo]
command = "dejo-mcp"

[mcp_servers.dejo.env]
DEJO_PROFILE = "/home/you/.dejo/profile-b"

Sites that need a sign-in: see Signing in.

FAQ

Does this work in the Codex IDE extension too?

Yes. The Codex CLI, the IDE extension and the ChatGPT desktop app share the same config.toml, so adding Dejo once makes it available in all of them.

Do I need an OpenAI key for Dejo?

No. Dejo never calls a model. Codex does the thinking; Dejo is the browser and the memory.

Make your agent remember.

Free and open source. Install it once, and every site your agent learns stays learned.

Get started