Skip to main content
The @output.ai/cli provides commands for initializing projects, running workflows, and AI-assisted development.

​
Installation

npm install -g @output.ai/cli

​
Quick Reference

CommandDescription
output initInitialize a new project
output devStart development services
output agents initInitialize AI agent configurations
output workflow planGenerate a workflow plan
output workflow generateGenerate a workflow from plan
output workflow listList available workflows
output workflow runExecute a workflow synchronously
output workflow startStart a workflow asynchronously
output workflow statusGet workflow execution status
output workflow outputGet workflow execution output
output workflow stopStop a workflow execution

​
Project Commands

​
output init

Initialize a new Output Framework workflow project. 
output init [folder-name]
​
folder-name
string
Name of the project folder to create
Flags:
FlagTypeDefaultDescription
--skip-envbooleanfalseSkip interactive environment variable configuration

​
output dev

Start all development services. 
output dev
This command starts:
ServiceURLDescription
Temporal UIhttp://localhost:8080Monitor and debug workflows
Temporal Serverlocalhost:7233gRPC endpoint
API Serverhttp://localhost:3001REST API for workflow execution
Worker—Processes workflows with auto-reload
PostgreSQLlocalhost:5432Temporal persistence
Redislocalhost:6379Caching layer
Flags:
FlagTypeDefaultDescription
--compose-file, -fstring—Path to custom docker-compose file
--no-watchbooleanfalseDisable file watching

​
output dev eject

Eject Docker Compose configuration for customization. 
output dev eject
Flags:
FlagTypeDefaultDescription
--output, -ostringdocker-compose.ymlOutput path
--force, -fbooleanfalseOverwrite existing file

​
Agent Commands

​
output agents init

Initialize agent configuration files for AI assistant integration. 
output agents init
This creates:
  • .outputai/ directory with agent and command configurations
  • .claude/ directory with symlinks for Claude Code integration
Flags:
FlagTypeDefaultDescription
--agent-providerstringclaude-codeAgent provider
--force, -fbooleanfalseOverwrite existing files

​
Workflow Commands

​
output workflow plan

Generate a workflow plan from a natural language description. 
output workflow plan
The command prompts for a workflow description, generates a plan using AI, and provides an interactive modification loop. Type ACCEPT to save the plan. Flags:
FlagTypeDefaultDescription
--description, -dstring—Workflow description (prompts if not provided)
--force-agent-file-writebooleanfalseForce overwrite agent templates
Example: 
output workflow plan --description "A workflow that summarizes news articles"

​
output workflow generate

Generate a new Output Framework workflow. 
output workflow generate <name>
​
name
string
required
Name of the workflow to generate
Flags:
FlagTypeDefaultDescription
--skeleton, -sbooleanfalseGenerate minimal skeleton without examples
--description, -dstring—Workflow description
--output-dir, -ostringworkflows/Output directory
--force, -fbooleanfalseOverwrite existing directory
--plan-file, -pstring—Path to plan file for AI-assisted implementation
Examples: 
# Generate skeleton workflow
output workflow generate my-workflow --skeleton

# Generate from a plan file
output workflow generate summarize --plan-file .outputai/plans/2025_01_15_summarize/PLAN.md

​
output workflow list

List available workflows from the catalog. 
output workflow list
Flags:
FlagTypeDefaultDescription
--format, -fstringlistOutput format: list, table, json
--detailed, -dbooleanfalseShow detailed information
--filterstring—Filter by name

​
output workflow run

Execute a workflow synchronously and wait for completion. 
output workflow run <workflowName>
​
workflowName
string
required
Name of the workflow to execute
Flags:
FlagTypeDefaultDescription
--input, -istring—JSON input or file path (required)
--task-queue, -qstring—Task queue name
--format, -fstringtextOutput format: json, text
Example: 
output workflow run simple --input '{"question": "who is ada lovelace?"}'

​
output workflow start

Start a workflow asynchronously without waiting for completion. 
output workflow start <workflowName>
​
workflowName
string
required
Name of the workflow to start
Flags:
FlagTypeDefaultDescription
--input, -istring—JSON input or file path (required)
--task-queue, -qstring—Task queue name
Returns the workflow ID immediately.

​
output workflow status

Get the status of a workflow execution. 
output workflow status <workflowId>
​
workflowId
string
required
The workflow execution ID
Flags:
FlagTypeDefaultDescription
--format, -fstringtextOutput format: json, text

​
output workflow output

Get the output of a completed workflow execution. 
output workflow output <workflowId>
​
workflowId
string
required
The workflow execution ID
Flags:
FlagTypeDefaultDescription
--format, -fstringtextOutput format: json, text

​
output workflow stop

Stop a running workflow execution. 
output workflow stop <workflowId>
​
workflowId
string
required
The workflow execution ID to stop

​
Environment Variables

The CLI respects the following environment variables:
VariableDescription
API_URLCustom API server URL (default: http://localhost:3001)
TEMPORAL_ADDRESSTemporal server address
TEMPORAL_NAMESPACETemporal namespace
ANTHROPIC_API_KEYAnthropic API key for Claude-based workflows
OPENAI_API_KEYOpenAI API key for GPT-based workflows