meta-cognition-parallel

# Meta-Cognition Parallel Analysis (Experimental)

> **Status:** Experimental | **Version:** 0.1.0
>
> This skill tests parallel three-layer cognitive analysis using `context: fork`.

## Concept

Instead of sequential analysis, this skill launches three parallel subagents - one for each cognitive layer - then synthesizes their results.

```
User Question
     │
     ▼
┌─────────────────────────────────────────────────────┐
│            meta-cognition-parallel                   │
│                  (Coordinator)                       │
└─────────────────────────────────────────────────────┘
     │
     ├─── Task(fork) ──► layer1-analyzer ──► L1 Result
     │                   (Language Mechanics)
     │
     ├─── Task(fork) ──► layer2-analyzer ──► L2 Result
     │                   (Design Choices)         ├── Parallel
     │                                            │
     └─── Task(fork) ──► layer3-analyzer ──► L3 Result
                         (Domain Constraints)
     │
     ▼
┌─────────────────────────────────────────────────────┐
│              Cross-Layer Synthesis                   │
│         (In main context with all results)          │
└─────────────────────────────────────────────────────┘
     │
     ▼
Domain-Correct Architectural Solution
```

## Usage

```
/meta-parallel <your Rust question>
```

**Example:**
```
/meta-parallel 我的交易系统报 E0382 错误，应该用 clone 吗？
```

## Execution Instructions

### Step 1: Parse User Query

Extract from `$ARGUMENTS`:
- The original question
- Any code snippets
- Domain hints (trading, web, embedded, etc.)

### Step 2: Launch Three Parallel Agents

**CRITICAL: Launch all three Tasks in a SINGLE message to enable parallel execution.**

```
Read agent files, then launch in parallel:

Task(
  subagent_type: "general-purpose",
  run_in_background: true,
  prompt: <content of agents/layer1-analyzer.md>
          + "\n\n## User Query\n" + $ARGUMENTS
)

Task(
  subagent_type: "general-purpose",
  run_in_background: true,
  prompt: