Lambda CLI – MCP Server and GPU Instance Manager

TL;DR

TL;DR: Lambda CLI is an unofficial Rust-based CLI and MCP server that lets you manage Lambda cloud GPU instances from the terminal or hook them into any MCP-compatible AI assistant like Claude Code.

Source and Accuracy Notes

⚠️ This section is MANDATORY. All links must be verified from actual source, not guessed.

• Project page: github.com/Strand-AI/lambda-cli — MUST read README
• Official Lambda site: cloud.lambda.ai — note this is an unofficial community project
• License: MIT (verified via Cargo.toml license = "MIT")
• HN launch thread: news.ycombinator.com/item?id=46621786
• npm package: npmjs.com/package/@strand-ai/lambda-mcp
• MCP registry: modelcontextprotocol.io

What Is Lambda CLI?

Lambda CLI is an unofficial community-built tool for managing GPU instances on Lambda Cloud. It ships in two forms:

• lambda — a direct terminal CLI for launching, stopping, and listing GPU instances
• lambda-mcp — an MCP server that exposes those same capabilities as MCP tools, letting any MCP-compatible AI assistant manage your infrastructure

The README describes it as:

A fast CLI and MCP server for managing Lambda cloud GPU instances.

The project is built in Rust with Clap for CLI argument parsing. Binaries are published to GitHub Releases and the package is distributed via npm (@strand-ai/lambda-mcp), Homebrew, and Cargo.

Installation

Homebrew (macOS / Linux)

brew install strand-ai/tap/lambda-cli

From source

cargo install --git https://github.com/Strand-AI/lambda-cli

Pre-built binaries

Download from the GitHub Releases page for Linux, macOS, and Windows.

Via npx (MCP server only)

npx -y @strand-ai/lambda-mcp

Authentication

You need a Lambda Cloud API key from the Lambda dashboard.

Option 1 — environment variable

export LAMBDA_API_KEY=your_key_here

Option 2 — secret manager command

export LAMBDA_API_KEY_COMMAND="your_command_here"

This executes the command at startup and uses its output as the API key. Works with 1Password, Pass, or any tool that can read secrets from the command line.

MCP Tools

The lambda-mcp MCP server exposes these tools:

| Tool | Description | |---|---| | start_instance | Launch a new GPU instance (auto-notifies if configured) | | stop_instance | Terminate a running instance | | list_running_instances | Show all running instances with status and connection details | | check_availability | Check if a specific GPU type is available |

Claude Code Setup

claude mcp add lambda -s user -e LAMBDA_API_KEY=your_key -- npx -y @strand-ai/lambda-mcp

Restart Claude Code after adding. Then you can prompt it to manage GPU infrastructure:

• “What GPUs are currently available on Lambda?”
• “Launch an H100 instance with my SSH key ‘macbook’”
• “Show me my running instances”
• “Check if any A100s are available”
• “Terminate instance i-abc123”

Notifications

Configure one or more notification targets and the MCP server sends alerts when instances become SSH-ready:

export LAMBDA_NOTIFY_SLACK_WEBHOOK="https://hooks.slack.com/services/..."
export LAMBDA_NOTIFY_DISCORD_WEBHOOK="https://discord.com/api/webhooks/..."
export LAMBDA_NOTIFY_TELEGRAM_BOT_TOKEN="..."   # + LAMBDA_NOTIFY_TELEGRAM_CHAT_ID

Development

# Build from source
cargo build

# Run tests
cargo test

# Run CLI directly
cargo run --bin lambda -- list

# Run MCP server directly
cargo run --bin lambda-mcp

FAQ

Q: Is this an official Lambda product? A: No. Lambda CLI is an unofficial community project, not affiliated with or endorsed by Lambda Labs.

Q: Does it work with all Lambda GPU instance types? A: The tool calls the Lambda Cloud API, which covers the instance types Lambda offers. The exact GPU inventory depends on Lambda Cloud region availability.

Q: Do I need a Lambda account to use this? A: Yes. You need an active Lambda Cloud account and an API key from their dashboard.

Q: Is the MCP server stable enough for production use? A: This is v0.5.3 as of writing. Check the GitHub releases for the latest status and any known issues.

Q: Can I use this with Cursor or other MCP-compatible editors? A: Yes. The README shows VS Code Copilot and Cursor install links alongside Claude Code.

Conclusion

Lambda CLI fills a gap if you want GPU infrastructure-as-code without clicking through the Lambda dashboard. The MCP server layer is the more interesting part — it means any AI assistant that speaks MCP can manage your GPU fleet alongside its usual tasks. Given that many developer workflows now bounce between a terminal and an AI pair programmer, having both the CLI and the MCP server in one binary is a pragmatic design.

The catch is that this is an unofficial, community-built tool (v0.5.3 at time of writing). Treat it accordingly: test against your actual Lambda usage before relying on it in anything sensitive.