implement-phases

# Phase Implementation Orchestration Skill

This skill provides the methodology for analyzing plan documents, determining optimal execution strategies, and coordinating phase implementation through sub-agents.

## Overview

When implementing phases from a plan document, the orchestrator must:
1. Extract and understand each phase's scope
2. Detect dependencies (explicit and implicit)
3. Choose optimal execution strategy
4. Coordinate sub-agents for implementation
5. Handle failures gracefully
6. Aggregate and report results

---

## CRITICAL: Mandatory Sub-Agent Requirement

**YOU MUST USE THE TASK TOOL TO SPAWN A SUB-AGENT FOR EVERY PHASE IMPLEMENTATION.**

This is a non-negotiable requirement. The orchestrator (main agent) is ONLY responsible for:
- Reading and parsing the plan document
- Analyzing dependencies
- Determining execution strategy
- Presenting the plan to the user
- Spawning Task() sub-agents for each phase
- Aggregating results after sub-agents complete

**The orchestrator MUST NOT:**
- Implement any phase directly in the main conversation
- Write code, create files, or make changes for any phase
- Skip sub-agent spawning for "simple" phases
- Combine multiple phases into a single implementation

**Why this matters:**
- Context isolation: Each sub-agent has fresh context, preventing saturation
- Parallelization: Independent phases can run in parallel via multiple Task() calls
- Failure isolation: A failed phase doesn't corrupt the main agent's state
- Results tracking: Sub-agents write structured results to coordination directory

**Correct pattern:**
```
# For each phase (or group of parallel phases), spawn Task() sub-agents
Task(
  subagent_type="general-purpose",
  prompt="[Phase implementation prompt with full spec and acceptance criteria]",
  description="Implement phase-1: [name]"
)
```

**WRONG pattern (never do this):**
```
# Never implement phases directly
Edit(file_path="src/feature.ts", ...)  # WRONG - orchestrator should not edit files
Wri