Task Tracker Workflow

Skill: Task Tracker Workflow (DAG Management)

This skill ensures that complex engineering tasks are broken down into manageable, trackable, and verifiable sub-tasks using Gemini's internal task tracking system.

The Protocol

Step 1: Initialize the DAG

For any task classified as "Epic" or "Massive", use tracker_add_task to create a parent task. Then, decompose the mission into specific sub-tasks (e.g., "Implement API", "Add Validation", "Write Tests").

Step 2: Establish Dependencies

Use the dependencies parameter in tracker_add_task to link sub-tasks. For example, "Write Tests" should depend on "Implement API". This prevents starting work before prerequisites are met.

Step 3: Progressive Execution

• Set Active: Use tracker_update_task to set a task's status to in_progress.
• Verify: Before marking a task as completed, run a specific verification command (e.g., test or lint).
• Update: Reflect progress in the tracker after every major sub-task completion.

Step 4: Visualizing State

Periodically use tracker_visualize to show the current progress tree to the user. This provides high-transparency and ensures the agent and user are aligned on the current state.

Best Practices

• Atomic Tasks: Keep sub-tasks small enough to be completed in 1-2 agent turns.
• Dependency First: Always resolve blocked tasks before moving to parallel tracks.
• Persistence: Since the tracker persists state outside the chat history, use it to resume long-running work across multiple sessions.