domrun 0.1.0

Python CLI and SDK helpers for Domrun Browser, MCP registries, and GPT/Codex automation workflows.

Domrun Python CLI and SDK

domrun is the Python distribution for Domrun Browser setup, diagnostics, and customer-owned MCP registry onboarding. It provides the same domrun command as the npm package, plus small Python helpers for backend integration tests.

Install

After the package is published to PyPI:

pip install domrun
domrun --help

With uv:

uv tool install domrun
domrun --help

Run without installing globally:

uvx domrun --help

For a project virtual environment:

uv pip install domrun

Configure a customer backend

domrun init \
  --registry-url https://mcp.customer.com \
  --automation-url https://automation.customer.com \
  --mcp-url https://mcp.customer.com/mcp \
  --user-key customer-acme:user_123 \
  --codex

This writes ~/.domrun/config.json. With --codex, it also writes:

• ~/.codex/config.toml with a domrun MCP server entry
• AGENTS.md in the current project with Domrun usage guidance for GPT/Codex agents

Useful commands

domrun browser open https://example.com
domrun connect "https://mcp.customer.com/v1/domrun/connect-sessions/cs_123"
domrun status
domrun browser status
domrun codex install --mcp-url https://mcp.customer.com/mcp
domrun mcp status
domrun doctor
domrun tools list

Python helper

from domrun import DomrunClient

client = DomrunClient(
    backend_url="https://mcp.customer.com",
    service_token="service-token",
)

session = client.create_connect_session(
    tenant_id="customer-acme",
    user_id="user_123",
    user_key="customer-acme:user_123",
    registry_url="https://mcp.customer.com",
    automation_url="https://automation.customer.com",
    browser_alias="Faizan's MacBook",
)

print(session["browser_connect_url"])

Publish to PyPI

The package is configured for public PyPI distribution as domrun.

Build and check locally:

uv run --with build python -m build
uv run --with twine python -m twine check dist/*

Publish after logging into the account that should own the package:

uv run --with twine python -m twine upload dist/*

For CI, use a PyPI trusted publisher or set TWINE_USERNAME=__token__ and TWINE_PASSWORD=<pypi-token>.