lambda_api

Minimal Python CLI + helpers for the Lambda Cloud API (https://cloud.lambda.ai/api/v1).

Install

Using the workspace venv (recommended):

• pip install -r requirements.txt
• pip install -e .

Auth

Set your API key:

• export LAMBDA_API_KEY=...

Or put it in a .env file in the project directory:

• create .env containing LAMBDA_API_KEY=...

The CLI will automatically load .env if present. You can also pass a specific file:

• lambda-api --dotenv /path/to/.env instances list --json

Commands

List running instances

• lambda-api instances list --json

List available images

List machine images you can reference in your launch JSON:

• lambda-api images list --json

To launch using a specific image, set one of the following in your launch JSON:

• By id: "image": {"id": "<IMAGE_ID>"}
• By family: "image": {"family": "<IMAGE_FAMILY>"}

Alert on instances running longer than 24 hours

Best-effort start time inference:

1. Prefer an instance tag containing an ISO8601 start time (default tag key: started-at).
2. Optionally fallback to scanning audit events (heuristic).

Example:

• lambda-api instances long-running --hours 24 --fail-on-findings --fallback-audit-events

Exit codes:

• 0: ok (or findings but --fail-on-findings not set)
• 1: findings present and --fail-on-findings set
• 2: error talking to API / config error

Discord bot integration (simple)

From a Discord bot, you can shell out and parse JSON:

• run lambda-api instances long-running --hours 24 --json
• if exit code is 1, post an alert with the returned long_running list

Notes on reliability

The Instances API responses shown in the public docs do not include a created/started timestamp. If you need rock-solid “age” tracking, the most reliable approach is to ensure instances are launched with a tag like:

• key: started-at
• value: 2025-12-17T12:34:56Z

This tool will use that tag when present.

Launch (start) an instance

Launch a new instance using the Lambda Cloud launch endpoint.

This CLI requires a JSON file containing the request body for POST /instance-operations/launch.

Example launch.json:

{
	"region_name": "us-west-2",
	"instance_type_name": "gpu_1x_a10",
	"ssh_key_names": ["my-ssh-key-name"],
	"quantity": 1
}

Run:

• lambda-api instances launch launch.json --json

Notes:

• The CLI automatically injects a started-at tag at launch time (unless you already provided one).
• Output is JSON (success envelope) and suitable for bot parsing.

Shutdown (terminate) an instance

Terminate an instance by its id:

• lambda-api instances shutdown <INSTANCE_ID> --json