Installation¶
This guide covers the one-line SuperQode installer and direct uv installation, including verification, optional runtime extras, and troubleshooting.
Safety note (OSS): Run the open-source SuperQode in a safe, controlled environment (sandbox, VM, or low-risk machine). This reduces the blast radius for testing workflows and agent-driven actions.
Quick Install¶
Run the hosted installer on macOS, Linux, or WSL:
curl -fsSL https://super-agentic.ai/superqode.sh | sh
It:
- uses an existing uv installation when available;
- otherwise announces and runs Astral's official uv installer;
- installs the latest SuperQode release from PyPI in an isolated tool environment;
- verifies both
superqode --versionandsq --version; and - never uses
sudo.
If you prefer to inspect downloaded scripts before running them:
curl -fsSL https://super-agentic.ai/superqode.sh \
-o superqode.sh
less superqode.sh
sh superqode.sh
Prefer to install it yourself?¶
The installer is the recommended path because it handles the uv bootstrap for you. If uv is already installed, the direct equivalent is:
uv tool install --upgrade --force superqode
Or run once without installing:
uvx superqode
Both are fully supported, so pick whichever fits your setup. See Installation Methods for the full comparison.
Verify the installed tool:
superqode --version
The installed package exposes two equivalent commands:
superqode harness list
sq harness list
sq is the short alias for every superqode command. The longer name remains the clearest choice for documentation, automation, and agents. For a one-off shorthand invocation without installing, use uvx --from superqode sq; uvx sq refers to a package named sq, not to SuperQode. If another application already owns sq on your PATH, continue using the unambiguous superqode command.
System Requirements¶
Local Model Hardware
The SuperQode CLI is lightweight, but local model serving is not. Running Ollama, LM Studio, MLX, vLLM, SGLang, DS4, or llama.cpp can use significant CPU, GPU, memory, battery, and cooling capacity. Use local models only on hardware that can safely support the selected model and context size.
Minimum Requirements¶
| Component | Requirement |
|---|---|
| Operating System | macOS 12+, Linux (Ubuntu 20.04+, Debian 11+), Windows 10+ (WSL2) |
| Python | 3.12 or higher |
| Memory | 4GB RAM minimum, 8GB recommended |
| Disk Space | 500MB for installation |
Python Version Check¶
python3 --version
# Should output: Python 3.12.x or higher
If you need to install Python 3.12+:
brew install python@3.12
sudo apt update
sudo apt install python3.12 python3.12-venv
sudo apt update
sudo apt install python3.12 python3.12-venv
Installation Methods¶
| Method | Best for | Needs uv preinstalled |
|---|---|---|
| One-line installer | Most users; first-time setup | No, it installs uv for you |
| uv directly | You already use uv and want to manage the install | Yes |
| uvx | Trying SuperQode without installing | Yes |
Method 1: One-line installer (recommended)¶
curl -fsSL https://super-agentic.ai/superqode.sh | sh
Run the same command later to upgrade. Uninstall with:
uv tool uninstall superqode
Method 2: uv directly¶
Use uv for an isolated tool installation managed separately from project environments.
# Install uv if needed
curl -LsSf https://astral.sh/uv/install.sh | sh
# Install SuperQode
uv tool install superqode
Method 3: uvx (No Persistent Install)¶
Run SuperQode directly through uv when you want a temporary command.
uvx superqode --version
Installing optional extras in one line¶
Set SUPERQODE_EXTRAS immediately before the installer. For example, install the Hugging Face Tau adapter:
curl -fsSL https://super-agentic.ai/superqode.sh |
SUPERQODE_EXTRAS=tau sh
Multiple extras use commas:
curl -fsSL https://super-agentic.ai/superqode.sh |
SUPERQODE_EXTRAS=tau,vendor-sdks sh
Release automation can install an exact published version with SUPERQODE_VERSION=0.2.68.
Installation for Developers¶
For contributors or those wanting the latest features:
# Clone the repository
git clone https://github.com/SuperagenticAI/superqode.git
cd superqode
# Install using uv (recommended for dev)
uv sync --extra dev --extra docs
# Verify installation
uv run superqode --version
Environment-Aware Extras¶
The TUI presents the normal packaged-product command and does not modify the environment until you run that command explicitly:
| Running from | Command shape |
|---|---|
uv tool install superqode | uv tool install "superqode[<extra>]" |
| SuperQode source checkout (contributors) | uv sync --extra <extra> |
Do not add SuperQode to an unrelated application's dependencies merely to enable a CLI integration.
To install the Codex, GitHub Copilot, Claude Agent, and Antigravity SDK runtimes together, use the optional vendor-sdks extra. It is intentionally excluded from the default installation because the vendor packages are large and many users need only one of them.
uv tool install "superqode[vendor-sdks]"
superqode runtime setup
For direct package installs such as MLX, SuperQode targets the running interpreter explicitly:
uv pip install --python /path/to/superqode/python "mlx-lm>=0.31.0,<0.32.0"
Post-Installation Setup¶
1. Initialize Configuration¶
# In your project directory, create a repo config
cd /path/to/your/project
superqode config init
This creates superqode.yaml in the current directory with local-first defaults for Ollama and qwen3:8b.
2. Set Up Local Models¶
For the local-first path, start Ollama and pull a starter model:
ollama pull qwen3:8b
Then run superqode, use :local init, :connect local, and select superqode.local.yaml with :harness superqode.local.yaml.
3. Set Up API Keys (BYOK Mode)¶
For cloud providers, set your API keys as environment variables:
export ANTHROPIC_API_KEY=sk-ant-...
export OPENAI_API_KEY=sk-...
export GOOGLE_API_KEY=...
# or
export GEMINI_API_KEY=...
Add to your shell profile (~/.bashrc, ~/.zshrc):
# SuperQode API Keys
export ANTHROPIC_API_KEY=sk-ant-...
export OPENAI_API_KEY=sk-...
export GOOGLE_API_KEY=...
export DEEPSEEK_API_KEY=...
export GROQ_API_KEY=...
3. Verify Provider Configuration¶
# Check authentication status
superqode auth info
# List available providers
superqode providers list
# Test a specific provider
superqode providers test anthropic
Optional Dependencies¶
Install an extra into the same environment that runs SuperQode:
uv tool install "superqode[<extra>]"
| Extra | Capability |
|---|---|
hf | Hugging Face Hub search, downloads, and accelerated transfers |
mlx | MLX and MLX-VLM local inference |
monty | Sandboxed Monty Python REPL |
semantic | CocoIndex semantic code search |
extras | Optional Exa Python client for external search integrations |
channels | Slack and Discord transports for the channel daemon |
sandbox-e2b | E2B cloud sandbox |
sandbox-modal | Modal cloud sandbox |
sandbox-daytona | Daytona cloud sandbox |
sandbox-cloud | E2B, Modal, and Daytona together |
web | Browser-hosted Textual TUI |
a2a | A2A HTTP server dependencies |
adk | Google Agent Development Kit runtime |
openai-agents | OpenAI Agents SDK runtime |
codex-sdk | OpenAI Codex SDK runtime |
copilot-sdk | GitHub Copilot SDK runtime |
claude-agent-sdk | Anthropic Claude Agent SDK runtime |
antigravity-sdk | Google Antigravity SDK runtime |
vendor-sdks | Codex, Copilot, Claude Agent, and Antigravity SDK runtimes |
deepagents | DeepAgents runtime |
pydanticai | PydanticAI runtime |
pydanticai-logfire | PydanticAI runtime with Logfire instrumentation |
rlm-code | RLM Code recursive harness backend |
tau | Hugging Face Tau Harness Protocol adapter and read-only TUI preset |
mem0 | Mem0 memory provider |
supermemory | Supermemory provider |
memory-providers | Mem0 and Supermemory together |
optimization | GEPA optimization engine |
observability | OpenTelemetry, MLflow, LangSmith, Logfire, and Arize Phoenix exporters |
The default install remains intentionally small. Install only the integrations used by the current repository or runtime.
Contributor and repository-maintenance extras are separate:
| Extra | Capability |
|---|---|
dev | Pytest, coverage, Ruff, MyPy, and pre-commit |
testing | CI test, security, and HTTP test dependencies |
linters | Bandit, Pylint, Flake8, Safety, and pip-audit |
ui-testing | Selenium and Playwright |
performance | Locust performance testing |
docs | MkDocs, Material for MkDocs, and documentation plugins |
For Local Models¶
# macOS
brew install ollama
# Linux
curl -fsSL https://ollama.com/install.sh | sh
# Start Ollama
ollama serve
# Pull a model
ollama pull qwen3:8b
Download from lmstudio.ai and install the desktop application.
uv pip install vllm
For ACP Agents¶
npm i -g opencode-ai
# Verify installation
opencode --version
Linting and Type Checking¶
Install language-specific linters for code analysis in harness runs:
uv pip install ruff mypy pyright
npm install -g eslint typescript
go install golang.org/x/lint/golint@latest
go install honnef.co/go/tools/cmd/staticcheck@latest
rustup component add clippy
Verify Installation¶
Basic Verification¶
# Check version
superqode --version
# View help
superqode --help
# Launch TUI
superqode
Full Verification¶
# Check all dependencies
superqode auth info
# List providers
superqode providers list
# List agents
superqode agents list
Expected Output¶
$ superqode --version
SuperQode v0.1.4
$ superqode auth info
โญโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฎ
โ Authentication Status โ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโค
โ Provider โ Status โ Model Access โ
โโโโโโโโโโโโโโโโโผโโโโโโโโโโโโผโโโโโโโโโโโโโโโโโโโโโโค
โ anthropic โ โ Valid โ <anthropic-model> โ
โ openai โ โ Valid โ <openai-model> โ
โ ollama โ โ Running โ qwen3:8b โ
โฐโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฏ
Troubleshooting¶
Common Issues¶
Python version too old
Error: requires Python 3.12+
Solution: Install Python 3.12 or higher using your package manager or pyenv.
# Using pyenv
pyenv install 3.12.0
pyenv global 3.12.0
Command not found after installation
Error: superqode: command not found
Solution: Ensure your PATH includes the ~/.local/bin directory used by uv.
export PATH="$HOME/.local/bin:$PATH"
source ~/.bashrc # or ~/.zshrc
Permission denied
Error: Permission denied during installation
Solution: Install with uv:
uv tool install superqode
SSL certificate errors
Error: SSL: CERTIFICATE_VERIFY_FAILED
Solution (macOS):
/Applications/Python\ 3.12/Install\ Certificates.command
Ollama connection refused
Error: Connection refused when using local models
Solution: Ensure Ollama is running:
ollama serve
Getting Help¶
If you encounter issues not covered here:
- Check the GitHub Issues
- Run with verbose logging:
superqode --verbose
Upgrading¶
Upgrade SuperQode¶
uv tool upgrade superqode
Check for Updates¶
# View current version
superqode --version
# Check PyPI for latest version
pip index versions superqode
Uninstalling¶
uv tool uninstall superqode
Remove Configuration¶
# Remove user configuration
rm -rf ~/.superqode.yaml
rm -rf ~/.superqode/
# Remove project configurations
rm -rf .superqode/
rm -f superqode.yaml
Next Steps¶
- Quick Start Guide - Get started in 5 minutes
- Your First Session - Complete walkthrough
- Configuration - Customize SuperQode