Workflow Automator

OpenClaw-native workflow automation engine — define, run, schedule, and manage task pipelines from the CLI.

Description

Workflow Automator is a lightweight, zero-dependency workflow engine built entirely in Bash. It lets AI agents and humans define multi-step task pipelines using a simplified YAML format, then run them sequentially with built-in logging, error handling, retry logic, conditional branching, and cron-based scheduling.

Unlike heavier orchestration tools that require Python, Node.js, or Docker, Workflow Automator runs anywhere Bash runs — your laptop, a VPS, a Raspberry Pi, or a CI/CD runner. It's designed to be the native automation backbone for OpenClaw deployments.

Use When

• Defining repeatable multi-step task pipelines (backup, deploy, monitor, report)
• Scheduling automated workflows via cron
• Running sequential commands with logging and error tracking
• Managing workflow execution history and status
• Generating workflow templates for common operations
• Exporting workflows to share with teammates or across machines
• Validating workflow files before execution
• Any automation task that would otherwise be a messy collection of shell scripts

Commands

| Command | Description | |---------|-------------| | init | Initialize a workflow project — creates directory structure, sample workflow file, and configuration | | run <file> | Execute a workflow file step-by-step with full logging, retry on failure, and status tracking | | validate <file> | Validate workflow file syntax — checks structure, required fields, and step definitions | | list | List all workflow files in the current project with name, step count, and last-run status | | status [run_id] | Show current running workflows or details of a specific run | | log [run_id] | View execution logs — latest run by default, or specify a run ID | | schedule <file> <cron> | Schedule a workflow to run on a cron expression (e.g., 0 2 * * * for daily at 2 AM) | | template <type> | Generate a workflow template: backup, deploy, monitor, report | | export <file> | Export a workflow as a self-contained shareable bundle (base64-encoded archive) | | history [n] | Show execution history — last 10 runs by default, or specify count |

Workflow File Format

Workflows use a simplified YAML format that can be parsed by Bash without external dependencies:

name: daily-backup
description: "Back up database and notify team. Use when you need workflow automator capabilities. Triggers on: workflow automator."
retry: 2
on_failure: notify

steps:
  - name: check-disk
    run: df -h | head -5
  - name: backup-db
    run: pg_dump mydb > /backups/backup-$(date +%Y%m%d).sql
    retry: 3
  - name: compress
    run: gzip /backups/backup-$(date +%Y%m%d).sql
    condition: test -f /backups/backup-$(date +%Y%m%d).sql
  - name: notify
    run: echo "Backup completed successfully at $(date)"

Supported Fields

• name (required): Workflow name identifier
• description: Human-readable description
• retry: Default retry count for all steps (default: 0)
• on_failure: Command to run if workflow fails
• steps: Ordered list of execution steps
  • name (required): Step identifier
  • run (required): Shell command to execute
  • retry: Per-step retry override
  • condition: Shell expression — step runs only if this returns 0
  • timeout: Maximum seconds for this step (default: 300)
  • on_failure: Per-step failure handler

Project Structure

After init, your workflow project looks like:

workflows/
├── .workflow/
│   ├── config          # Project configuration
│   ├── runs/           # Execution history and logs
│   └── schedules/      # Cron schedule definitions
├── templates/          # Generated templates
└── example.yml         # Sample workflow

Features

• Zero dependencies — pure Bash, no Python/Node/Ruby required
• Step-by-step execution — each step logged with timestamps and exit codes
• Error retry — configurable retry count per step or globally
• Conditional steps — skip steps based on shell conditions
• Cron scheduling — native crontab integration for recurring workflows
• Execution history — full audit trail of every run
• Templates — quick-start templates for backup, deploy, monitor, report
• Export/share — bundle workflows for sharing across machines
• Colored output — clear visual feedback during execution
• Parallel-safe — lock files prevent concurrent runs of the same workflow

Integration with OpenClaw

The agent can use Workflow Automator to:

1. Automate repetitive tasks — define once, run forever
2. Schedule maintenance — backups, cleanups, health checks via cron
3. Build deployment pipelines — multi-step deploy with rollback
4. Monitor systems — periodic checks with alerting on failure
5. Generate reports — scheduled data collection and formatting

Script

The main script is located at:

scripts/main.sh

Run commands via: bash scripts/main.sh <command> [args]

💬 Feedback & Feature Requests: https://bytesagain.com/feedback Powered by BytesAgain | bytesagain.com