sima-cli - SiMa Developer Portal CLI Tool

Documentation: English | 한국어 | 日本語 | 繁體中文 | Українська

sima-cli is the command-line interface for SiMa developer workflows. Use it to authenticate, set up SDK containers, update DevKits, install packages, download artifacts, and access Model Zoo and App Zoo content.

Documentation

The full command reference is generated as Markdown under docs/sima-cli.

Use the generated docs for detailed options, arguments, subcommands, and full help text:

• Complete command reference
• Top-level command help

Installation

For most users, install the latest official release from the public installer URL for your operating system.

Linux, macOS, and DevKit

Run the installer from a terminal:

curl -fsSL https://artifacts.neat.sima.ai/sima-cli/linux-mac.sh | bash

After installation, open a new terminal or reload your shell profile, then verify the install:

sima-cli --version

Windows PowerShell

Download and run the Windows installer from PowerShell:

Invoke-WebRequest https://artifacts.neat.sima.ai/sima-cli/windows.bat -OutFile windows.bat
.\windows.bat

After installation, open a new Command Prompt or PowerShell window, then verify the install:

sima-cli --version

Advanced: choose a branch or release

Use install.py only when you need to choose a specific tested branch build or release instead of installing the latest official PyPI release.

On Linux, macOS, or DevKit:

curl -fsSL https://artifacts.neat.sima.ai/sima-cli/install.py -o sima-cli-install.py
python3 sima-cli-install.py

Install a specific branch or release:

python3 sima-cli-install.py feature/my-branch latest
python3 sima-cli-install.py v2.1.6 latest

On Windows PowerShell:

Invoke-WebRequest https://artifacts.neat.sima.ai/sima-cli/install.py -OutFile sima-cli-install.py
python .\sima-cli-install.py

To install a specific branch or release:

python .\sima-cli-install.py feature/my-branch latest
python .\sima-cli-install.py v2.1.6 latest

Release tags such as v2.1.6 install from public PyPI. Branch names install tested artifacts from artifacts.neat.sima.ai/sima-cli.

Public PyPI releases can also be installed directly:

pip install sima-cli

Quick Start

sima-cli --help
sima-cli --version
sima-cli login

Use --internal or SIMA_CLI_INTERNAL=1 when internal Artifactory resources are required:

sima-cli --internal login
SIMA_CLI_INTERNAL=1 sima-cli install -v 2.1.2 tools/model-compiler/amd64

Full-system SWU updates prefer RAM-backed /tmp when available memory can hold both the bundle and its extraction, plus a reserve of at least 512 MiB or 10% of total RAM (whichever is larger). Otherwise they stage in /data, with NVMe as a further fallback if /data lacks space. RAM staging is temporary; interrupted updates retained there do not survive a reboot.

Public-mirror downloads, models download, and neat artifacts split files of at least 64 MiB into four byte ranges when supported by the server. Models and Neat artifacts share a limit of four active download connections across files and ranges, including single-connection fallbacks. Ranges must match the expected size and an object validator or the expected SHA-256 before the file is published. Models retain signed-URL refresh and cancellation handling. Existing public-mirror partial downloads continue sequentially; servers without range support use a single connection.

Common Workflows

Set up SDK containers:

sima-cli sdk setup
sima-cli sdk neat

When pairing with sdk setup --devkit, a new DevKit host key is automatically recorded in the SDK user's ~/.ssh/known_hosts, including during unattended setup or after SDK container recreation. Subsequent pairing preserves saved keys and rejects a changed key at the same address and SSH port. After a device reimage or key rotation, verify the replacement fingerprint before explicitly updating that entry. For environments requiring verified trust before the first connection, provision verified host keys in the SDK user's known_hosts.

Install Model Compiler from package metadata:

# amd64 host
sima-cli install -v 2.1.2 tools/model-compiler/amd64
# arm64 host
sima-cli install -v 2.1.2 tools/model-compiler/arm64

Download or install Neat artifacts:

sima-cli neat download core main
sima-cli install --neat core main

Update a DevKit:

sima-cli update -v 2.1.1 -y

Explore Model Zoo and App Zoo content:

sima-cli modelzoo list
sima-cli appzoo list

Top-Level Commands

Development

Install development dependencies and run tests:

pip install -e ".[dev]"
python -m pytest tests/unit

Regenerate CLI documentation:

python scripts/generate_cli_markdown_docs.py

build.sh also regenerates the command docs before building the package.

Requirements

• Python 3.8 or newer
• Docker for SDK container workflows
• DevKit connectivity for device update, serial, network, and boot-image workflows

For command-specific prerequisites, see the generated command reference.

Support

For issues and feature requests, use the sima-cli GitHub repository or contact the SiMa.ai development team.

Remote browser certificate help

SDK setup stages a verified public mkcert CA certificate alongside the SDK HTTPS certificate. SDK images with browser certificate help can offer that file for download on the sign-in page, with instructions for the browser computer. Only the public certificate is staged; the mkcert private key stays on the host. If the issuing CA cannot be verified, including a plain self-signed fallback, no CA download is published. Re-running setup refreshes the staged file even when the existing HTTPS certificate does not need to be reissued.