swe-swarm-analyze

## ⚠️ WORKFLOW INITIALIZATION

**If starting a new session**, first read workflow initialization:
```
mcp__plugin_swe_serena__read_memory("WF_INIT")
```
Follow WF_INIT instructions before executing this skill.

---

# Swarm Analyze Skill

Deep codebase analysis using Decentralized Autonomous Agents (DAA).

## When to Use

- Large codebases (1000+ files)
- Complex multi-module projects
- When detailed DOM_* and SYS_* memories are needed
- Feature onboarding with full analysis mode

## MCP Requirements

**Required (one of):**
- `ruv-swarm` MCP (preferred for DAA learning)
- `claude-flow` MCP (alternative)

**Fallback:** Sequential analysis if no swarm MCP available

## Agent Types

| Agent ID | Purpose | Cognitive Pattern |
|----------|---------|-------------------|
| config-analyzer | Parse config files | convergent |
| architecture-mapper | Detect layers | systems |
| pattern-detector | Find conventions | lateral |
| domain-extractor | Extract domains | divergent |
| system-finder | Identify systems | systems |
| test-analyzer | Test patterns | critical |
| import-tracer | Dependency graph | convergent |
| convention-learner | Style detection | adaptive |
| file-indexer | File inventory | convergent |
| synthesizer | Compile results | systems |

## Process

### Step 1: Initialize Swarm

**⚠️ CRITICAL: RUV-Swarm has TWO separate agent pools - choose ONE pattern:**

| Pattern | Agent Creation | Execution | Use When |
|---------|---------------|-----------|----------|
| **Swarm** | `agent_spawn` | `task_orchestrate` | Parallel task execution |
| **DAA** | `daa_agent_create` | `daa_workflow_execute` | Learning/adaptation needed |

```javascript
// Option A: RUV-Swarm Task Orchestration (faster, no learning)
if (mcp_available("ruv-swarm") && !needsLearning) {
  mcp__ruv-swarm__swarm_init({ topology: "mesh", strategy: "balanced", maxAgents: 10 });
}

// Option B: RUV-Swarm DAA Workflow (slower, with learning)
if (mcp_available("ruv-swarm") && needsLearning) {
  mcp__ruv-s