CLI Commands Reference

Complete reference for all ModelKnife command-line interface commands

💡 Access Control: Commands marked with 🔒 require developer access, and ⚠️ require admin access. Use mk team status to check your access level.

CLI Command Groups

ML Workflow Management

Infrastructure Management

Configuration & Setup

Team & Access Control

Development Tools

Pipeline Commands (mk p) 🔒

Manage ML workflows and pipeline deployments across SageMaker, Glue, and Step Functions

mk p deploy

Deploy ML pipeline modules with unified orchestration.

mk p deploy [OPTIONS]
Option Description Default
--base-path The folder path of the configuration file Current directory
--profile, -p The Profile name of the compose file None
--skip-build, -s Skip the build step False
--with-services Deploy services before deploying pipeline modules False

mk p run

Execute ML pipeline or specific modules.

mk p run [OPTIONS]
Option Description Default
--modules, -m Comma-separated list of modules to run All modules
--profile, -p The Profile name of the compose file None
--base-path Path to the configuration file Current directory

mk p show

Display detailed information about the pipeline configuration.

mk p show [OPTIONS]
Option Description Default
--profile, -p The Profile name of the compose file None
--base-path Path to the configuration file directory Current directory
--detailed, -d Show detailed information including resolved configurations False

mk p status

Check real-time execution status for pipeline modules. The command aggregates Step Functions execution details and summarizes module-level status across SageMaker, Glue, and other AWS services.

mk p status [OPTIONS]
Option Description Default
--name The specific pipeline name to check status for Current pipeline
-n, --execution Show the n-th last execution (1 = latest, 2 = second-to-last, etc.) 1 (latest)
--profile, -p The Profile name of the compose file None
--base-path Path to the configuration file directory Current directory
Usage Examples 
# Show latest execution status (default)
mk p status

# Show status for the second-to-last execution
mk p status -n 2

# Specify pipeline by name (when outside project directory)
mk p status --name my-pipeline

# Use a specific profile
mk p status -p prod

Output includes overall execution details and a single "Module Status" table:

Pipeline: my-pipeline
📊 Latest Execution: 2025-09-02-16-25-35
🏃 Overall Status: SUCCEEDED
⏰ Started: 2025-09-02 16:25:35
🔗 Monitor: AWS Console

📋 Module Status:
+---------------------+-----------+---------------------+----------+
| Module Name         | Status    | Start Time          | Duration |
+---------------------+-----------+---------------------+----------+
| feature-engineering | Succeeded | 2025-09-02 16:25:35 | 5m 0s    |
| model-training      | Failed    | 2025-09-02 16:30:36 | 5m 0s    |
+---------------------+-----------+---------------------+----------+

Columns:

  • Module Name: The pipeline module.
  • Status: Aggregated status (Running, Succeeded, Failed, etc.).
  • Start Time: Module start time (local timezone where available).
  • Duration: Elapsed time or completion duration.

mk p visualize / mk p viz

Generate an interactive visualization of the pipeline structure and dependencies.

mk p visualize [OPTIONS]
mk p viz [OPTIONS]  # Short alias
Option Description Default
--profile, -p The Profile name of the compose file None
--base-path Path to the configuration file directory Current directory
--output, -o Output file path for the visualization Auto-generated
--no-open Don't automatically open the visualization in browser False

mk p delete

Delete deployed pipeline resources from AWS.

mk p delete [OPTIONS]
Option Description Default
--name The specific pipeline name to delete Current pipeline
--profile, -p The Profile name of the compose file None
--base-path Path to the configuration file directory Current directory
--confirm Skip confirmation prompt False

mk p version

Show version information for deployed pipelines.

mk p version [OPTIONS]
Option Description Default
--list List all deployed pipeline versions False
--name Show version info for specific pipeline name Current pipeline
--profile, -p The Profile name of the compose file None
--base-path Path to the configuration file directory Current directory

mk p runs

Show execution history for pipelines.

mk p runs [OPTIONS]
Option Description Default
--name Pipeline name to show execution history for Current pipeline
--limit Number of recent executions to show 10
--profile, -p The Profile name of the compose file None
--base-path Path to the configuration file directory Current directory
Usage Examples 
# Show last 10 executions (default)
mk p runs

# Show last 5 executions
mk p runs --limit 5

# Show executions for a specific pipeline name
mk p runs --name my-pipeline

# With a specific profile
mk p runs -p prod --limit 20

mk p stop

Stop running pipeline executions.

mk p stop [OPTIONS]
Option Description Default
--module Stop specific module execution All modules
--execution-arn Stop specific execution by ARN None
--profile, -p The Profile name of the compose file None
--base-path Path to the configuration file directory Current directory
--confirm Skip confirmation prompt False

Pipeline Scheduling Commands

Configure automated pipeline execution schedules using EventBridge rules.

mk p schedule set

Create automated pipeline execution schedule.

mk p schedule set --cron "EXPRESSION" [OPTIONS]
Option Description Example
--cron Cron expression for schedule (required) "0 9 * * 1-5"
--timezone Timezone for the schedule Europe/Zurich
--description Optional description for the schedule "Daily ML training"
--profile, -p The Profile name of the compose file prod
--base-path Path to the configuration file directory ./config
Schedule Examples 
# Weekdays 9 AM UTC
mk p schedule set --cron "0 9 * * 1-5"

# Daily 00:10 Zurich time
mk p schedule set --cron "10 0 * * *" --timezone Europe/Zurich

# Daily 2 AM Beijing time with description
mk p schedule set --cron "0 2 * * *" --timezone Asia/Shanghai --description "Daily ML training"

mk p schedule show

Show the current schedule for a pipeline.

mk p schedule show [OPTIONS]

mk p schedule remove

Remove the schedule for a pipeline.

mk p schedule remove [OPTIONS]

mk p schedule list

List all scheduled pipelines in the current AWS account.

mk p schedule list [OPTIONS]

mk p schedule disable / enable

Disable or enable the schedule for a pipeline without removing it.

mk p schedule disable [OPTIONS]
mk p schedule enable [OPTIONS]

Service Commands (mk s) 🔒

Deploy and manage AWS infrastructure services (DynamoDB, Lambda, API Gateway, etc.)

mk s deploy

Deploy AWS infrastructure services defined in the configuration.

mk s deploy [OPTIONS]
Option Description Default
--service, -s Deploy specific services (comma-separated) All services
--base-path The folder path of the configuration file Current directory
--profile, -p Configuration profile None
--force Force redeploy all services regardless of changes False
--verbose Show detailed deployment output including build steps False

mk s status

Show deployment and operational status of infrastructure services.

mk s status [OPTIONS]
Option Description Default
--service, -s Show status for specific services (comma-separated) All services
--base-path Path to configuration file Current directory
--profile, -p Configuration profile None
--details Show detailed outputs and resource information False

mk s validate

Validate service configurations without deploying.

mk s validate [OPTIONS]

mk s delete

Delete deployed infrastructure services from AWS.

mk s delete [OPTIONS]
Option Description Default
--service, -s Delete specific services (comma-separated) All services
--confirm Skip confirmation prompt False

mk s types

List all available service types with descriptions.

mk s types

mk s rollback

Rollback services to previous deployment state.

mk s rollback [OPTIONS]

mk s template

Generate service configuration templates.

mk s template [OPTIONS]
Option Description Example
--type Service type to generate template for lambda_function
--name Service name for the template my_service
--interactive, -i Interactive mode to customize configuration False

mk s cache

Manage service deployment cache.

mk s cache [OPTIONS]
Option Description Options
--action Cache action to perform stats, clean, clear
--max-age Maximum age in days for cache cleanup 7

Configuration Commands (mk conf)

AWS profile configuration and system settings

mk conf show

Display current AWS profile configuration and account information.

mk conf show [OPTIONS]

Shows the current AWS profile being used by ModelKnife, account ID, region, and configuration source (file, environment variable, or default).

mk conf set

Set AWS profile for ModelKnife operations (persists across sessions).

mk conf set PROFILE_NAME
Argument Description Required
PROFILE_NAME AWS profile name to use for ModelKnife operations Yes
Usage Examples 
# Set production profile
mk conf set production

# Set development profile
mk conf set development

# Verify the setting
mk conf show

mk conf unset

Remove AWS profile configuration (uses environment variable or default AWS profile).

mk conf unset

After unsetting, ModelKnife will use the MLKNIFE_AWS_PROFILE environment variable or fall back to the default AWS profile.

Configuration Priority

ModelKnife checks for AWS profile configuration in this order:

  1. Config file - Set via mk conf set
  2. Environment variable - MLKNIFE_AWS_PROFILE
  3. Default - AWS default profile or environment credentials

Team Management (mk team)

Team access management and user permissions

mk team status

Check your current access level and team configuration.

mk team status [OPTIONS]

mk team list-users

List all team members in ModelKnife groups.

mk team list-users [OPTIONS]
Option Description Options
--group List users in specific group only mlknife-developers, mlknife-admin

mk team check-access

Check access permissions for different operations.

mk team check-access [OPTIONS]
Option Description Options
--operation Type of operation to check access for basic, admin

mk team setup ⚠️

Create IAM groups and policies for team access.

mk team setup [OPTIONS]

mk team add-user ⚠️

Add a user to a ModelKnife team group.

mk team add-user [OPTIONS]
Option Description Required
--user IAM username to add to group Yes
--group Group to add user to Yes (mlknife-developers or mlknife-admin)

mk team remove-user ⚠️

Remove a user from a ModelKnife team group.

mk team remove-user [OPTIONS]
Option Description Required
--user IAM username to remove from group Yes
--group Group to remove user from Yes (mlknife-developers or mlknife-admin)
--confirm Skip confirmation prompt No

Setup Commands (mk setup)

Automated setup wizard for team onboarding

mk setup init ⚠️

One-command setup for new teams - generates IAM configuration files and creates standardized AWS resources with account ID-based naming for team collaboration.

mk setup init [OPTIONS]

This command:

  • Generates IAM configuration files with standardized role naming using account ID
  • Creates IAM roles, groups, and policies using configuration-based approach
  • Sets up AWS resources (S3 buckets, EventBridge permissions, etc.)
  • Adds the current user to the admin group
  • Enables seamless team collaboration through shared role names
Option Description Default
--admin-user Username to add as admin Auto-detected
--skip-confirm Skip confirmation prompts False
--interactive Enable interactive mode True
--non-interactive Disable interactive prompts False

mk setup status

Check setup health and configuration status.

mk setup status [OPTIONS]

mk setup fix ⚠️

Fix common setup issues automatically.

mk setup fix [OPTIONS]

mk setup clean ⚠️

Remove all ModelKnife setup and AWS resources.

mk setup clean [OPTIONS]
Option Description Default
--confirm Skip confirmation prompts False

mk setup reset ⚠️

Reset user configuration to system defaults (admin only).

mk setup reset [OPTIONS]

Clears all user customizations and resets configuration to system defaults. Useful for removing custom policies and groups while keeping the base system configuration.

Option Description Default
--confirm Skip confirmation prompts False

Format Commands (mk fmt)

Format YAML configuration files with consistent style

mk fmt

Format YAML configuration files with consistent indentation and style.

mk fmt [OPTIONS]
Option Description Default
--profile, -p Profile name to format (e.g., dev, prod) mlknife-compose.yaml
--base-path Base path to search for configuration files Current directory
--indent, -i Number of spaces for indentation 2
--check Check if formatting is needed without making changes False
--diff Show diff of changes that would be made False
--all Format all mlknife-compose*.yaml files False
--mode Formatting mode auto (basic, reindent, auto)

Show Commands (mk show)

Display pipeline and service configuration overview

mk show

Display comprehensive overview of pipeline and service configurations.

mk show [OPTIONS]
Option Description Default
--profile, -p The Profile name of the compose file None
--base-path Path to the folder containing the configuration file Current directory
--detailed, -d Show detailed information including descriptions False

Access Control Summary

Understanding command permissions and requirements

📖 Public Commands (No Access Control)

Anyone can run these commands:

  • mk fmt - Format YAML configuration files
  • mk show - Display configuration overview
  • mk team status - View team status
  • mk team check-access - Check permissions
  • mk team list-users - List team members
  • mk setup status - Check setup status

🔒 Developer Access Required

Must be in mlknife-developers or mlknife-admin group:

  • All mk s commands (service management)
  • All mk p commands (pipeline management)
  • mk conf show - View configuration
  • mk conf import - Import configuration

⚠️ Admin Access Required

Must be in mlknife-admin group only:

  • mk team setup, mk team add-user, mk team remove-user
  • mk setup init, mk setup fix, mk setup clean, mk setup reset

Note: Configuration commands (mk conf) are now focused on AWS profile management and are available to all users.

💡 Getting Access: Use mk team status to check your current access level. Contact your admin to be added to the appropriate group using mk team add-user.