LMA CLI — Command-Line Interface for AWS Live Meeting Assistant

LMA CLI — Command-Line Interface for AWS Live Meeting Assistant

Overview

The LMA CLI (lma-cli) provides a rich command-line interface for building, deploying, and managing AWS Live Meeting Assistant. It uses the LMA SDK under the hood and provides beautiful terminal output via Rich.

Installation

# From the project root — install both SDK and CLI
pip install -e lib/lma_sdk
pip install -e lib/lma_cli_pkg

# Or use make
make setup-cli

After installation, the lma command (and alias lma-cli) is available:

lma --help

Commands

lma publish — Build & Upload Artifacts

Packages all LMA sub-stacks, uploads artifacts to S3, and generates a deployable CloudFormation template URL.

# Standard build and publish (bucket auto-generated from AWS account ID)
lma publish

# With custom bucket and prefix, and region
lma publish --bucket-basename my-artifacts --prefix lma --region us-east-1

# Force full rebuild (skip change detection)
lma publish --force

# Public artifacts (for shared deployments)
lma publish --public

Options:

Option	Description
--source-dir PATH	Path to LMA project root (default: .)
--bucket-basename TEXT	S3 bucket basename (auto-generated from account ID if omitted)
--prefix TEXT	S3 key prefix (default: lma)
--region TEXT	AWS region (default: from AWS CLI profile)
--public	Make artifacts publicly readable
--force	Force full rebuild (delete checksums)
--version TEXT	Override version string
--no-validate	Skip CloudFormation template validation
-v, --verbose	Enable verbose build output

Features:

• Rich progress output with per-stack status
• SHA256-based change detection (skips unchanged stacks)
• Colourized summary panel with template URL and console URL

lma deploy — Deploy/Update Stack

Deploy or update the LMA CloudFormation stack. If no template source is specified, deploys from the public published LMA template for the current region.

# Deploy from public template (simplest — new stack)
lma deploy --stack-name MyLMA --admin-email user@example.com --wait

# Deploy from public template (create new stack or update existing stack)
lma deploy --stack-name MyLMA --wait

# Deploy from local code (build, publish, then deploy)
lma deploy --stack-name MyLMA --from-code . --admin-email user@example.com --wait

# Deploy from local code with custom bucket
lma deploy --stack-name MyLMA --from-code . --bucket-basename my-artifacts --wait

# Deploy from specific S3 template URL
lma deploy --stack-name MyLMA --template-url https://s3.us-east-1.amazonaws.com/bucket/lma-main.yaml

# Deploy with parameter overrides
lma deploy --stack-name MyLMA -p AdminEmail=admin@example.com -p InstallDemoMode=true --wait

# Deploy from local template file
lma deploy --stack-name MyLMA --template-file /tmp/lma-main.yaml --wait

Options:

Option	Description
--stack-name TEXT	CloudFormation stack name (default: LMA)
--admin-email TEXT	Admin user email (required for new stacks)
--from-code DIRECTORY	Build and publish from local source before deploying
--template-url TEXT	S3 URL for CloudFormation template
--template-file FILE	Path to local CloudFormation template file
-p, --parameter KEY=VALUE	Parameter override (can be repeated)
--wait	Wait for stack operation to complete with event streaming
--no-rollback	Disable rollback on stack creation failure
--role-arn TEXT	CloudFormation service role ARN
--timeout INTEGER	Max wait time in minutes (default: 120)
--bucket-basename TEXT	S3 bucket basename (used with --from-code)
--prefix TEXT	S3 key prefix (default: lma, used with --from-code)
--public	Make S3 artifacts publicly readable (used with --from-code)
--clean-build	Force full rebuild (used with --from-code)
--no-validate-template	Skip CloudFormation template validation (used with --from-code)

Template Resolution:

• If --from-code is specified, builds and publishes artifacts first, then deploys the resulting template
• If --template-url is specified, uses that URL directly
• If --template-file is specified, uses the local file
• If none specified, auto-selects the public LMA template for the current region:
  • us-east-1, us-west-2, ap-southeast-2 are supported

Smart Features:

• Auto-detects if the stack has an operation in progress and switches to monitoring mode
• Streams CloudFormation events in real-time when --wait is used
• Validates that --admin-email is provided for new stack creation
• Shows next steps and important outputs upon completion

lma status — Stack Status

Show current CloudFormation stack status, parameters, and outputs.

lma status
lma status --stack-name MyLMA

lma outputs — Stack Outputs

Show stack outputs (CloudFront URL, AppSync endpoint, Cognito pool, etc.).

lma outputs
lma outputs --json
lma outputs --stack-name MyLMA

lma delete — Delete Stack

Delete the LMA stack and all its resources.

lma delete                    # Interactive confirmation
lma delete --yes              # Skip confirmation
lma delete --stack-name MyLMA --yes

lma logs — CloudWatch Logs

View CloudWatch logs for Lambda functions and services.

# List available log groups
lma logs --list

# View logs from a specific group
lma logs /LMA/lambda/FetchTranscript

# Partial name matching
lma logs FetchTranscript --since 60 --limit 200

lma check-prereqs — Prerequisite Check

Verify all publish prerequisites are installed (Docker, SAM CLI, Node.js, etc.).

lma check-prereqs

lma list-stacks — List Sub-Stacks

List all publishable LMA sub-stacks with their package types.

lma list-stacks

lma vp — Virtual Participant Operations

Create, fetch, end, and list Virtual Participants directly from the CLI. These commands talk to the LMA AppSync API and the VP scheduler Step Function via SigV4-signed AWS calls (no Cognito login required — your AWS credentials are used).

# Create + launch a VP, waiting until it leaves INITIALIZING
lma vp create --name "Weekly sync" --platform ZOOM --id 1234567890

# Create with a password and a custom display name; emit JSON
lma vp create --name "Standup" --platform TEAMS --id 99999 \
              --password "abc" --user-name "lma-bot" --json

# Get a single VP row
lma vp get --id <vp-id>

# End a running VP
lma vp end --id <vp-id> --reason "Manual cleanup"

# List all VPs visible to the calling IAM principal
lma vp list
lma vp list --json

Common options for every lma vp subcommand:

Option	Description
--stack-name TEXT	CloudFormation stack name (env: LMA_STACK_NAME)
--region TEXT	AWS region (env: AWS_DEFAULT_REGION)
--json	Emit JSON instead of formatted text

lma vp create extra options:

Option	Description
--name TEXT	Meeting name (required)
--platform [ZOOM|TEAMS|CHIME|WEBEX]	Meeting platform (required, case-insensitive)
--id TEXT	Meeting ID (required)
--password TEXT	Meeting password (default: empty)
--user-name TEXT	Display name the scribe reports (default: lma-cli@lma)
--wait / --no-wait	Poll until the VP leaves INITIALIZING (default: --wait)
--timeout FLOAT	Max seconds to wait for launch (default: 120)

The underlying SDK methods are documented under client.vp.

lma load — Load Simulator (plugin)

The lma load subcommand tree is provided by the LMA Load Simulator package (utilities/load-simulator/). It registers itself with the lma CLI through the lma_cli.plugins entry-point group, so once both packages are installed alongside each other it appears as a first-class subcommand of lma.

# Install the load simulator (in addition to lma-sdk + lma-cli)
pip install -e utilities/load-simulator

# Or with optional driver extras (websocket audio resampling, upload pipeline)
pip install -e 'utilities/load-simulator[all]'

# Verify
lma load --help

Available scenarios (each tagged with a --run-id for deterministic teardown):

Subcommand	Purpose
lma load concurrent	Drive N simultaneous meetings via --driver kinesis|upload|websocket|vp
lma load backfill	Fabricate N historical meetings spread over the last Y days (synthetic or real Cognito users for RBAC-at-scale testing)
lma load rbac	Provision N synthetic Cognito users + latency-sweep listCalls / getCallCount
lma load cleanup	Delete every synthetic resource (meetings, VPs, users, S3 orphans) for a given --target-run-id (or *)
lma load stack-info	Print the CloudFormation resources the simulator resolved — handy for debugging

The simulator can also be invoked standalone as lma-load <subcommand> if you have not installed lma-cli.

For the full reference (driver matrix, cost/quota guardrails, observability dashboard, synthetic-user provisioning, safety rails, and example fixtures), see the LMA Load Simulator README.

CLI Plugins

The lma CLI auto-discovers any installed Python package that declares a Click command under the lma_cli.plugins entry-point group. This is how lma load ... is contributed by the Load Simulator package; the same mechanism can be used for any third-party command tree.

# pyproject.toml of your plugin package
[project.entry-points."lma_cli.plugins"]
mything = "my_pkg.cli:my_command"   # registers `lma mything ...`

Plugin failures are logged but never crash the CLI — if a plugin fails to import, its commands simply don’t appear under lma --help.

Global Options

lma --region us-west-2 status                  # Override AWS region
lma --profile prod outputs                     # Use specific AWS profile
lma --stack-name MyLMA status                  # Override stack name
lma -v publish --source-dir . --region us-east-1  # Verbose logging

Environment Variables

Variable	Description	Default
LMA_STACK_NAME	CloudFormation stack name	LMA
AWS_DEFAULT_REGION	AWS region	us-east-1
AWS_PROFILE	AWS CLI profile	(default chain)

Typical Workflows

Quick Deploy (from public template)

# Deploy the latest published LMA — auto-selects template for your region
lma deploy --stack-name MyLMA --admin-email user@example.com --wait

Developer Workflow (from local code)

# 1. Check prerequisites
lma check-prereqs

# 2. Build and deploy from local source code in one step
lma deploy --stack-name MyLMA --from-code . --admin-email user@example.com --wait

# 3. Get the CloudFront URL
lma outputs

# 4. View logs
lma logs --list
lma logs FetchTranscript --since 30

Manual Publish + Deploy

# 1. Publish artifacts to S3
lma publish --source-dir . --region us-east-1

# 2. Deploy using the published template URL
lma deploy --stack-name MyLMA --template-url <template-url> --wait

# 3. Monitor status
lma status

Monitor an In-Progress Deployment

# If a deploy is already running, this auto-detects and monitors it
lma deploy --stack-name MyLMA --wait

Future Commands (Phase 2+)

Command	Description	Phase
lma meetings list	List meetings	2
lma meetings get <id>	Get meeting details	2
lma transcript get <id>	Get transcript	2
lma ask "<question>"	Ask meeting assistant	4
lma stream start	Stream audio	4

Virtual Participant join/leave/list is already shipped — see lma vp. Load testing is provided by the lma load plugin.

Architecture

The CLI is built with Click and Rich, following the same patterns as the IDP CLI:

lma_cli/
├── cli.py           # Main entry point, Click group + plugin loader
├── formatters.py    # Rich output helpers (tables, panels, colours)
└── commands/
    ├── publish.py   # publish, check-prereqs, list-stacks
    ├── stack.py     # status, outputs, deploy, delete, logs
    └── vp.py        # vp create / get / end / list

External plugins (e.g. utilities/load-simulator/lma_load/cli.py, which contributes lma load ...) are auto-registered via the lma_cli.plugins entry-point group — see CLI Plugins.

All AWS operations go through the LMA SDK — the CLI only handles UX/formatting.