Installation¶

This guide covers how to install RLM Code, its optional dependencies, and how to verify your installation.

System Requirements¶

Install uv¶

We recommend uv as the primary way to install and manage RLM Code.

# macOS / Linux
curl -LsSf https://astral.sh/uv/install.sh | sh

# Or with Homebrew
brew install uv

uv python install 3.12

Standard Installation¶

uv tool install "rlm-code[tui,llm-all]"

uv pip install "rlm-code[tui,llm-all]"

pip install "rlm-code[tui,llm-all]"

This installs the core package, the TUI, and all LLM provider clients:

Minimal Installation¶

If you only need one LLM provider:

# Core + TUI + Anthropic only
uv tool install "rlm-code[tui,anthropic]"

# Core + TUI + OpenAI only
uv tool install "rlm-code[tui,openai]"

# Core + TUI + Gemini only
uv tool install "rlm-code[tui,gemini]"

pip install "rlm-code[tui,anthropic]"
pip install "rlm-code[tui,openai]"
pip install "rlm-code[tui,gemini]"

Development Installation¶

For contributors or those who want to run from source:

git clone https://github.com/SuperagenticAI/rlm-code.git
cd rlm-code
uv sync --all-extras
uv run pytest

git clone https://github.com/SuperagenticAI/rlm-code.git
cd rlm-code
python -m venv .venv
source .venv/bin/activate
pip install -e ".[dev,tui,llm-all]"

The dev extra installs:

Optional Dependencies¶

Runtime Backend Requirements¶

Pick at least one secure backend before running serious experiments.

Observability Integrations¶

If you installed with uv tool install, use uv tool install --with to add extras, or reinstall with additional extras:

uv tool install "rlm-code[tui,llm-all,mlflow]"

uv tool install "rlm-code[tui,llm-all,mlflow]"

# If using uv tool, reinstall with the extra packages:
uv tool install "rlm-code[tui,llm-all]" --with opentelemetry-api --with opentelemetry-sdk --with opentelemetry-exporter-otlp-proto-grpc

uv tool install "rlm-code[tui,llm-all]" --with langsmith

uv tool install "rlm-code[tui,llm-all]" --with langfuse

uv tool install "rlm-code[tui,llm-all]" --with logfire

Framework Adapters¶

uv tool install "rlm-code[tui,llm-all,frameworks]"

MCP WebSocket Transport¶

uv tool install "rlm-code[tui,llm-all,mcp-ws]"

Adds websockets >= 15.0.1 for WebSocket-based MCP server transport.

Docker Runtime¶

Docker is used as a sandbox runtime for isolated code execution. No pip install is needed, but Docker must be available on the system:

# macOS
brew install --cask docker

# Linux (Ubuntu/Debian)
sudo apt-get install docker.io

# Verify Docker is running
docker info

Monty Backend¶

Monty is an optional secure backend for pure RLM execution:

pip install pydantic-monty

In TUI:

/sandbox backend monty

Apple Container Runtime (macOS, Experimental)¶

container --version
container system status

In TUI:

/sandbox apple on
/sandbox use apple-container

Documentation¶

uv tool install "rlm-code[docs]"

Installs mkdocs, mkdocs-material, mkdocstrings, and mkdocs-minify-plugin for building these docs locally.

Full Installation (Everything)¶

To install with all optional dependencies at once:

uv tool install "rlm-code[tui,llm-all,mlflow,frameworks,mcp-ws]"

pip install "rlm-code[tui,llm-all,mlflow,frameworks,mcp-ws]"

Verification¶

After installation, verify that everything works:

Check the version¶

rlm-code --version

Check sandbox runtimes¶

Launch the TUI and run the sandbox doctor:

rlm-code

Then in the TUI:

/sandbox doctor

This runs diagnostics on all available sandbox runtimes (local, Docker, Apple Container, Modal, E2B, Daytona) and reports their health status.

Check observability sinks¶

/rlm observability

This displays the status of all configured observability sinks (Local JSONL, MLflow, OpenTelemetry, LangSmith, LangFuse, Logfire).

Verify Python environment¶

python -c "import rlm_code; print(rlm_code.__version__)"

Upgrading¶

uv tool upgrade rlm-code

pip install --upgrade "rlm-code[tui,llm-all]"

Uninstalling¶

uv tool uninstall rlm-code

pip uninstall rlm-code

Troubleshooting¶

uv tool install "rlm-code[tui,llm-all]"

# macOS
open -a Docker

# Linux
sudo systemctl start docker

uv tool install --force "rlm-code[tui,llm-all]"