complete-process

# Complete Process Orchestrator

Auto-chain: Transform → Validate → Optimize

## Memory Integration (START - MANDATORY)

Before starting the orchestration pipeline, load session memory:

```
# Step 1: Create memory directory (permission-free)
Bash(command="mkdir -p .claude/pseudo-code-prompting")

# Step 2: Load memory files (permission-free)
Read(file_path=".claude/pseudo-code-prompting/activeContext.md")
Read(file_path=".claude/pseudo-code-prompting/patterns.md")
Read(file_path=".claude/pseudo-code-prompting/progress.md")
```

**Memory is passed through pipeline**:
- Transform step receives memory context from patterns.md
- Validation step uses patterns.md + progress.md for learned validations
- Optimization step applies patterns.md + progress.md for effective optimizations
- Memory is updated at END of pipeline with complete results

## MUST FOLLOW THIS 3-STEP WORKFLOW

When user invokes `/complete-process` or says "use complete process", run these 3 skills in sequence:

## Welcome Message and Menu System

When users invoke the plugin using trigger phrases, you MUST display a welcome message with an interactive menu for skill selection.

**See:** [references/welcome-menu-system.md](references/welcome-menu-system.md) for complete menu behavior and routing logic.

## CRITICAL IMPLEMENTATION REQUIREMENTS

### 1. Always Use Skill Tool for Sub-Skills

**MANDATORY**: When executing transformation, validation, or optimization steps, you MUST use the Skill tool to invoke the respective skills. NEVER handle these directly or inline.

```text
Skill(pseudo-code-prompting:prompt-structurer, args=user_query)
Output: transformed_pseudo_code

Display: "✓ Step 1/3 complete | Tokens: [N]
         Transformed to pseudo-code"
```

### 2. Context Window Optimization (MANDATORY)

To minimize token usage, you MUST remove intermediate outputs from the conversation flow:

**Keep Only**:

- Original user query (input)
- Final optimized output

**Remove/Don't Store**:

- Transform step ou