Workflow Reference

55+ automated workflows that make your agent smarter every day.

The Three Tiers

AIBrain workflows are organized into three tiers that work together as a system:

Tier 2 powers Tier 3. The Contradiction Detector can only find contradictions if Memory Hygiene keeps the brain clean, Correction Learner captures your feedback, and User Model tracks your preferences. Without the background work, the intelligence features fail.

Quick Start

# Enable the recommended starter set
$ aibrain workflows enable --recommended

# Sync to your OS scheduler
$ aibrain workflows sync

# Preview what would be created (optional)
$ aibrain workflows sync --dry-run

The recommended set enables 7 agent improvement workflows, 5 personal intelligence workflows, and 2 utility workflows (daily planner + weekly review).

Productivity

Research & Intelligence

Communication

Developer & DevOps

Self-Learning

These are the Tier 2 and Tier 3 workflows that make AIBrain fundamentally different. They run in the background and compound over time.

System

Job Search & Career

Recommended Starting Set

Start with these 5

1. correction_learner — the single most important workflow. Every correction becomes a permanent improvement.
2. memory_hygiene — keeps the brain fast and clean as it grows.
3. self_model_refresh — the agent knows what it's good at and where it struggles.
4. decision_journal — start tracking decisions now. In 3 months you'll have judgment quality data.
5. weekly_wins — surfaces accomplishments you already forgot about.

After 2 weeks, add

• contradiction_detector — enough decisions stored for useful detection
• learning_path — see which skills you're building
• life_dashboard — single view of everything that matters

After 1 month, add

• burnout_detector — enough data to spot quality patterns
• pattern_matcher — enough task history to find recurring themes
• idea_incubator — enough ideas stored to find connections

Scheduling & Cron

Workflows run on your OS scheduler. No daemon, no Docker, no always-running process.

# Create native scheduled tasks from enabled workflows
$ aibrain workflows sync

# Preview changes without applying
$ aibrain workflows sync --dry-run

This creates:

• Linux: entries in your user crontab
• Windows: tasks via schtasks
• macOS: launchctl plist files

You can also define schedules manually in scheduled_jobs.json:

{
  "weather_briefing": {
    "enabled": true,
    "cron": "30 6 * * *",
    "script": "workflows/weather_briefing.py"
  },
  "memory_consolidation": {
    "enabled": true,
    "cron": "0 3 * * 0",
    "script": "workflows/memory_consolidation.py"
  }
}

All Tier 1 utility workflows use free APIs (Open-Meteo, CoinGecko, ArXiv, Hacker News, Reddit). No paid API keys required.

Dependencies

When you enable a Tier 3 workflow, its Tier 2 dependencies auto-enable:

$ aibrain workflows enable contradiction_detector

Enabled: contradiction_detector
Auto-enabled dependencies:
  correction_learner
  memory_hygiene
  user_model

If you disable a Tier 2 workflow that Tier 3 workflows depend on, you'll see a warning. The dependent workflows will still run but may produce less accurate results.

# View the full dependency map
$ aibrain workflows deps

# View dependencies for a specific workflow
$ aibrain workflows deps contradiction_detector

Custom Workflows

Create custom workflows by subclassing SkillWorkflow or UserWorkflow:

# my_workflow.py
from aibrain.workflows import UserWorkflow

class MyCustomWorkflow(UserWorkflow):
    name = "my_custom_workflow"
    schedule = "0 9 * * *"  # 9am daily
    description = "My custom daily workflow"

    def run(self, brain):
        # Read from the brain
        memories = brain.search("project status")

        # Do your work
        summary = self.process(memories)

        # Write back to the brain
        brain.store(
            content=summary,
            memory_type="project",
            importance=7
        )

You can also create workflows from the dashboard using the Visual Workflow Builder (drag-and-drop node editor) or via chat with the /create command.

All Commands

FAQ

Do workflows use API credits or make network calls?
No. All Tier 2 and Tier 3 workflows operate on local SQLite data only. Some Tier 1 utility workflows (RSS, GitHub, weather) make HTTP requests but these are optional and clearly marked.

Will workflows slow down my system?
No. Each workflow runs for a few seconds at most, on its own schedule. There's no always-running daemon. Your OS scheduler triggers them like any other cron job.

Can I create custom workflows?
Yes. Subclass SkillWorkflow or UserWorkflow, implement run(), and register it. Or use the Visual Workflow Builder in the dashboard.

What happens if I disable a dependency?
Dependent Tier 3 workflows will still run but may produce less accurate results. You'll see a warning when disabling a dependency.

Does this work air-gapped?
Yes. All Tier 2 and Tier 3 workflows are fully air-gap safe. Zero network required.