quick-start_skill

Overview

This skill provides an interactive quick-start onboarding for the MCP Task Orchestrator. It detects whether your workspace is empty or already has tracked work and guides you through plan mode, persistent tracking, and how MCP items map to execution. The walkthrough is hands-on: create a tutorial project, advance items, and inspect cross-session continuity.

How this skill works

The skill calls the workspace health check to choose a Fresh-Start or Orientation path. In the Fresh-Start path it demonstrates plan mode, creates an atomic work tree (container + child items + dependencies), and walks through starting and completing tasks. In the Orientation path it builds a condensed dashboard from context and item queries, explains blocked and stalled work, and recommends concrete next actions.

When to use it

• When you say: get started, quick start, or onboard me
• First-time setup or trying the task orchestrator tutorial
• You want to see how plan mode and MCP items interact
• You need a dashboard-style orientation of active/blocked work
• To learn note schemas and gating transitions

Best practices

• Run get_context() first to detect workspace state before making changes
• Use create_work_tree for atomic project creation to avoid partial state
• Rely on plan files for design decisions and MCP items for execution state
• Use note schemas to enforce documentation gates before advancing items
• Call get_next_item() to discover recommended next work across sessions

Example use cases

• Create a tutorial project for a web feature and see design→implement→test lifecycle
• Onboard a new agent to an existing workspace by showing the health dashboard and recommendations
• Recover mid-session work by checking get_context() and plan file for continuity
• Unblock stalled items by reviewing missing notes with get_context(itemId=...) and upserting notes
• Define a feature-implementation schema to require requirements/design notes before work starts