Pipelines and Scheduling

#Concepts

#Projects

A project in t0ggles Crew is a local code folder linked to a t0ggles board and project. All pipelines in a project run against the same folder and board context.

By default, only one pipeline per project runs at a time (serial execution) to prevent file system conflicts. You can increase this in project settings.

#Pipeline Templates

Pipeline templates are reusable blueprints that define an agent's behavior. A template includes:

• Agent identity - name, avatar, bot user ID, model
• Prompt - instructions for the agent (supports template variables)
• Schedule - when and how often to run
• Timeout - maximum execution time

Templates can be shared across projects. When you assign a template to a project, you can override any field for that specific assignment.

#Pipeline Assignments

A pipeline assignment is a project's reference to a pipeline template. It can either use the template as-is or override specific fields (prompt, schedule, timeout, agent identity) for that project.

You can also create fully custom pipelines without a template.

#Prompt Variables

Pipeline prompts support template variables that get replaced at runtime:

Variable	Description
{{project.name}}	Project name
{{project.board}}	t0ggles board name
{{project.boardId}}	t0ggles board ID
{{agent.name}}	Agent display name
{{agent.botId}}	Agent bot user ID
{{timestamp}}	Current ISO timestamp
{{date}}	Current date (YYYY-MM-DD)

#Runs

A run is a single execution of a pipeline. Each run tracks:

• Start and completion time
• Duration
• Status: running, success, error, timeout, or cancelled
• Full output log
• Summary (generated by the agent)

The last 100 runs per pipeline are stored for history and debugging.

For real-world prompt examples, see Pipeline Examples.

#Schedule Modes

t0ggles Crew supports five scheduling modes for controlling when pipelines run.

#Manual

Pipelines only run when you click "Run Now". No automatic scheduling.

Best for: one-off tasks, testing prompts, on-demand operations.

#Interval

Run every N minutes on a repeating cycle.

• Interval: 15, 30, 60, 120, 240, 480, 720, or 1440 minutes
• Offset: Start at a specific minute of the hour (:00, :15, :30, :45)
• Active days: Optionally restrict to specific days of the week
• Active window: Optionally restrict to specific hours (e.g., 9 AM - 6 PM)

Best for: regular check-ins, periodic code reviews, recurring maintenance.

#Fixed Times

Run at specific times on specific days.

Each entry specifies a time (HH:MM in 24-hour format) and optional days of the week. You can add multiple time entries.

Example: Run at 09:00 on weekdays and 14:00 on weekends.

Best for: daily standups, end-of-day reports, scheduled deployments.

#After Pipeline

Run after another pipeline completes. Chains pipelines together.

• Trigger pipeline: The pipeline that triggers this one
• Delay: Wait N minutes after the trigger completes (0-60)
• On success only: Optionally only trigger if the predecessor succeeds

Best for: code review after feature implementation, testing after code changes, deployment after review.

#Auto

Automatically triggers when new tasks are assigned to the agent's bot user on the t0ggles board, or when the agent is mentioned in a task comment.

Requirements:

• The project must be connected to a t0ggles board
• The agent must have a bot user ID configured

The app polls every 5 minutes. When it finds tasks assigned to the agent in a non-done status, it automatically enqueues a run.

Comment mentions: If someone writes @AgentName in a task comment, the agent picks up that task on the next poll - even if the task isn't assigned to it. The agent reads the comment thread, takes the requested action, and posts a reply. Once it replies, the mention is considered handled and won't re-trigger.

Best for: fully autonomous agents that pick up tasks and respond to follow-up requests without manual intervention.