tk-workflow

iheanyi's avatarfrom iheanyi

Show the Tasuku task workflow guide. Use when user says /tk:workflow or asks how to use Tasuku, task workflow, or needs help with task management.

⭐18stars🔀0forks📁View on GitHub🕐Updated Jan 10, 2026⬇️2downloads

When & Why to Use This Skill

The Tasuku Workflow Guide is a specialized Claude skill designed for agent-first task management. It provides a structured framework for managing task lifecycles, tracking progress, and coordinating multi-agent workflows. By offering clear commands for starting, blocking, and completing tasks, it ensures seamless project execution and comprehensive documentation of insights and decisions, making it an essential tool for maintaining productivity in AI-driven environments.

Use Cases

  • Onboarding users and AI agents to the Tasuku ecosystem with standardized task management protocols and command references.
  • Streamlining multi-agent collaboration by implementing task claiming and release mechanisms to prevent workflow overlaps and resource conflicts.
  • Enhancing project transparency through systematic status updates (ready, in_progress, blocked, done) and real-time progress tracking.
  • Managing complex task dependencies and blockers in a structured, agent-oriented development environment to ensure continuous delivery.
  • Building a persistent knowledge base by documenting insights, decisions, and task notes directly within the project workflow.
nametk-workflow
descriptionShow the Tasuku task workflow guide. Use when user says /tk:workflow or asks how to use Tasuku, task workflow, or needs help with task management.

Tasuku Workflow Guide

Tasuku is an agent-first task management system. Here's how to use it effectively.

Task Lifecycle

ready → in_progress → done
          ↓
       blocked → (unblock) → ready

Common Operations

Starting Work

1. tk_list status=ready     # See what's available
2. tk_start id=task-id      # Claim the task
3. tk_timer_start id=task-id # Optional: track time

Finishing Work

1. tk_timer_stop id=task-id  # Stop timer if running
2. tk_done id=task-id        # Mark complete
3. tk_note task_id note="What was done"  # Optional: add notes

When Blocked

1. tk_block id=task-id blocked_by=["other-task"]
2. Work on something else from ready queue
3. tk_unblock id=task-id  # When blocker is resolved

Recording Context

tk_learn insight="Important discovery"   # Save learnings
tk_decide id=name chose=X over=[Y,Z] because="reason"  # Document decisions
tk_note task_id=id note="Details"        # Add task notes

Multi-Agent Coordination

When multiple agents work in parallel:

tk_claim id=task-id agent=my-name   # Exclusive claim
tk_release id=task-id               # Release when done

Claims expire after 2 hours if not renewed.

Best Practices

  1. Keep tasks small - One task = one PR
  2. Update status immediately - Don't batch updates
  3. Record learnings - Future sessions benefit
  4. Document decisions - Capture the "why"
  5. Use priorities - Critical (0) to Backlog (4)