orchestrate

Orchestrate Command

Coordinate complex tasks through sequential delegation to specialist agents.

Arguments

• --use-deep-dive: Use existing deep-dive context to skip or accelerate exploration phase
• <task description>: The task to orchestrate

State Initialization

FIRST: Initialize orchestration state by running:

# Check for --use-deep-dive flag
USE_DEEP_DIVE=false
TASK_ARGS="$ARGUMENTS"
if [[ "$ARGUMENTS" == *"--use-deep-dive"* ]]; then
  USE_DEEP_DIVE=true
  TASK_ARGS=$(echo "$ARGUMENTS" | sed 's/--use-deep-dive//' | xargs)
fi

bash ${CLAUDE_PLUGIN_ROOT}/scripts/init-orchestration.sh "$TASK_ARGS"

This creates .claude/orchestration.local.md to track:

• Current phase and iteration
• Gate results for each phase
• Agent actions and timestamps

Prompt Refinement (Pre-Phase)

Before beginning orchestration, ensure the task is well-defined:

1. Check Task Clarity: Does "$ARGUMENTS" specify:

   • What needs to be changed?
   • Where in the codebase?
   • What problem it solves?

2. If Vague: Ask ONE clarifying question before proceeding

   • Provide options when possible
   • Reference prompt-refinement skill for guidance

3. If Clear: Transform into structured format:

   • Goal: One-sentence outcome
   • Description: What and why (2-3 sentences)
   • Actions: Concrete steps
   • Constraints: Non-negotiable limits
   • Assumptions: Things believed true that, if false, would change the approach

4. Classify task complexity using the task-classification skill tiers (Trivial / Exploratory / Implementation / Complex / Research). Note: task_complexity is the task-classification tier, NOT complexipy code/cognitive complexity.

5. Detect explicit written-report request: independently of the complexity tier, determine whether the user explicitly asked for a written report, investigation guide, or planning document (e.g., "write me a report", "give me an investigation guide", "produce a planning doc"). Set REPORT_REQUESTED_FLAG to true or false accordingly.