Tencent/teamai-cli

Recommendation – Install teamai‑cli if you need a command‑line interface for Tencent TeamAI

The teamai‑cli binary provides documented commands for authentication, model deployment, project management and log retrieval. Using the CLI avoids manual construction of HTTP requests and reduces the likelihood of breaking changes when the public API evolves. The tool is open‑source, runs on Linux, macOS and Windows, and can be called from scripts or CI pipelines.

---

1. What is teamai‑cli and what problems does it address?

teamai‑cli is a command‑line client for Tencent TeamAI, a service that provides hosting, serving, and monitoring of machine‑learning models. The CLI translates concise commands into authenticated calls to the TeamAI REST API, removing the need for hand‑crafted requests.

Core purpose

Main features

• Authentication flow – teamai login stores a short‑lived JWT token in ~/.teamai/config.json; tokens can be refreshed automatically.
• Project and model management – list, create, delete and update projects; upload model artifacts; roll back deployments.
• Deployment orchestration – teamai deploy packages code, uploads the artifact and starts a serving instance in a single step.
• Monitoring utilities – fetch health checks, resource utilisation and streaming logs without leaving the shell.
• Plug‑in system – community‑contributed commands can be placed in ~/.teamai/plugins.
• Cross‑platform binary – distributed as an npm package that includes pre‑compiled binaries for Linux, macOS and Windows, ensuring identical behaviour across operating systems.

---

2. Installation on Linux, macOS and Windows

The primary distribution channel is npm, the package manager for Node.js. The package contains the compiled binary, so no local compilation is required. A Docker image (tencent/teamai-cli) is also published for environments where Node.js cannot be installed.

Prerequisites

Note: The CLI is available globally; some regions with strict internet controls may need a proxy or VPN to reach the npm registry.

Linux (Ubuntu/Debian, Fedora, Arch, etc.)

1. Install Node.js – the NodeSource script works on most distributions:

   curl -fsSL https://deb.nodesource.com/setup_18.x | sudo -E bash -
   sudo apt-get install -y nodejs   # use the appropriate package manager for your distro
   

2. Verify the installation:

   node -v   # should show v18.x or later
   npm -v    # should show 8.x or later
   

3. Install the CLI globally:

   sudo npm install -g @tencent/teamai-cli
   

4. Confirm the binary is on the PATH:

   teamai --version
   

   If the command is not found, add the global npm bin directory (e.g., /usr/local/bin) to PATH.

macOS (Intel and Apple Silicon)

1. Install Node.js via Homebrew:

   brew update
   brew install node
   

2. Install the CLI:

   npm install -g @tencent/teamai-cli
   

3. Verify the installation:

   teamai --version
   

   Permission errors can be avoided by using Homebrew’s Node, which writes to a user‑writable directory; sudo is not required.

Windows (10, 11, Server)

1. Install Node.js – download the LTS MSI installer from https://nodejs.org and run it. The installer adds node and npm to the system PATH.

2. Open PowerShell as Administrator and run:

   npm install -g @tencent/teamai-cli
   

3. Check the installation:

   teamai --version
   

   If the command is not recognized, restart the terminal or add %APPDATA%\npm to the system PATH.

Optional: Docker wrapper

When Node.js cannot be installed, the CLI can be executed inside Docker:

docker run --rm -it \
  -v "$HOME/.teamai:/root/.teamai" \
  -v "$(pwd):/workspace" \
  tencent/teamai-cli:latest \
  teamai <command> [options]

The volume mount preserves authentication tokens between container runs, making the approach suitable for CI environments that already use Docker.

---

3. Core commands and typical workflows

All commands read the default configuration file at ~/.teamai/config.json unless overridden with --config. The examples below assume a project named my‑project.

3.1 Authentication

teamai login

• Opens a browser (or prints a URL) where you sign in with your Tencent Cloud credentials.
• On success, a JWT token is saved to ~/.teamai/config.json under accessToken.
• Tokens expire after 24 hours; refresh with teamai refresh-token.

For non‑interactive environments (e.g., CI), set TEAMAI_TOKEN to a pre‑generated API token from the TeamAI console.

3.2 Listing projects and models

teamai project list
teamai model list --project my-project

• project list prints a table with projectId, name and createdAt.
• model list shows registered models, their version tags and timestamps.

Add --json to any list command to receive machine‑readable output, which is useful in scripts.

3.3 Deploying a model

Assume you have a compressed artifact model.tar.gz and a deployment descriptor deploy.yaml.

teamai deploy \
  --project my-project \
  --model-name sentiment-v2 \
  --artifact ./model.tar.gz \
  --config ./deploy.yaml

The command uploads the artifact, creates a version entry and starts a rollout. Progress is shown as a spinner; on completion a deployment ID is printed.

3.4 Monitoring and logs

teamai status --deployment <deployment-id>
teamai logs --deployment <deployment-id> --follow

• status reports health checks, CPU/memory usage and the current serving state.
• logs streams container logs; --follow keeps the connection open for real‑time output.

Both commands accept --json for programmatic consumption.

---

4. Troubleshooting common issues

When a problem is not covered here, consult the project’s issue tracker on GitHub for similar reports or open a new issue with the command output and environment details.

---

5. Contributing to the project

The source code resides at https://github.com/Tencent/teamai-cli. Contributions are accepted via pull requests to the main branch.

1. Fork the repository and clone it locally.
2. **