team-tracking-orchestrate

team-tracking-orchestrate

Orchestration has two modes — plan (decide what goes on the board, decompose, dispatch) and supervise (keep dispatched specialists on track). Each is its own skill. This entry exists for back-compat and to route you to the right one.

Pick by where you are

Where you are	What to do
Turning a goal/PRD/bug into board state. Reading the existing board. Choosing hierarchy + priority. Decomposing a task into pipeline subtasks. Promoting parents to Todo. Briefing specialists for dispatch.	Spawn the team-tracking-planner subagent. Hand it the PRD/intent and the project name; it returns a structured dispatch_list. Keeps board reading and decomposition out of your context.
At least one specialist is in flight. Listening on the event log. Reading checkpoints. Posting nudges/questions. Recovering stale locks. Acting on Blocked / Done events.	Load team-tracking-supervise.

The two phases interleave

In a real session you alternate: spawn the planner → dispatch what it returned → supervise → a specialist releases as Done or Blocked → spawn the planner again for the next pipeline stage (or to re-plan around the blocker) → dispatch → supervise. Planning lives in the subagent; supervision lives in your session.

Other skills you'll reach for

• team-tracking-planner — the subagent you spawn for every plan / re-plan. Owns team-tracking-plan internally so you don't have to load it.
• team-tracking-execute — the protocol the specialist runs. Read it once so you know what they're doing on the other side of the dispatch.
• team-tracking-usage — tool reference: the eleven MCP tools, the four ticket types, the unified event log, the lock state machine.
• team-tracking-obsidian-kanban — adapter quirks if your project uses the obsidian-kanban backend (file layout, card-eligibility rule, sub-bullet rendering, do-not-edit ## Children).