coding-gateway is a lightweight launcher for running Codex, Claude Code, Gemini CLI, OpenCode, and GitHub Copilot CLI through Databricks.

• Python 3.12+ — install with uv (uv.astral.sh)
• npm if tool CLIs need to be installed automatically

uv tool install git+https://github.com/databricks/coding-gateway

Just run the tool you want:

coding-gateway codex      # OpenAI Codex
coding-gateway claude     # Claude Code
coding-gateway gemini     # Gemini CLI
coding-gateway opencode   # OpenCode
coding-gateway copilot    # GitHub Copilot CLI

On first launch, coding-gateway will prompt for your Databricks workspace URL, authenticate, and configure that tool automatically. Subsequent launches go straight to the agent.

Pass flags directly to the underlying tool:

coding-gateway claude -r          # resume last session
coding-gateway codex --full-auto

All agents route through Databricks AI Gateway using your workspace credentials — no API keys required.

To configure all tools at once:

coding-gateway configure

coding-gateway configure mcp

Add Databricks MCP servers to installed MCP-capable tools: Codex, Claude Code, Gemini CLI, OpenCode, and GitHub Copilot CLI. Options are shown in this order:

• Discovered external MCP connections
• Databricks SQL
• Managed Databricks MCPs (Vector Search, UC Functions, etc.)
• Custom MCP server URL

Discovered external MCP connections are listed directly. MCP auth uses a Databricks token that coding-gateway sets when launching each tool.

coding-gateway manages these files:

Existing files are backed up before being overwritten. coding-gateway revert restores backups.

• Databricks AI Gateway overview
• Databricks AI Gateway coding agent integration
• Databricks CLI authentication
• Monitor AI Gateway usage

Contributions are welcome.

git clone https://github.com/databricks/coding-gateway
cd coding-gateway
uv sync

1. Create a feature branch off main.

2. Make your changes — keep them scoped to the requested behavior.

3. Run the test suite before pushing:

   uv run pytest          # unit tests
   uv run ruff check .    # lint

4. For end-to-end testing against a real workspace:

   CODING_GATEWAY_TEST_WORKSPACE=<db_workspace_url> uv run pytest tests/test_e2e.py -v

5. Open a pull request against main.

• Add src/coding_tool_gateway/agents/<name>.py with at least write_tool_config, launch, default_model, and validate_cmd.
• Register it in src/coding_tool_gateway/agents/__init__.py.
• Add focused tests under tests/.

Please report security vulnerabilities to security@databricks.com rather than opening a public issue.

See LICENSE.md and NOTICE.md.