Meta Prompt Architecture

# Meta-Prompt Architecture Implementation Plan

> **For agentic workers:** REQUIRED: Use superpowers:subagent-driven-development (if subagents available) or superpowers:executing-plans to implement this plan. Steps use checkbox (`- [ ]`) syntax for tracking.

**Goal:** Replace scaffold's hard-coded prompt pipeline with a meta-prompt + knowledge base architecture supporting three methodology tiers (Deep Domain Modeling, MVP, Custom).

**Architecture:** Meta-prompts declare step intent (30-80 lines each). Knowledge base provides domain expertise. At runtime, the CLI assembles meta-prompt + knowledge + project context + user instructions into a single prompt, and the AI generates and executes the working prompt. Methodology presets control which steps are active and at what depth (1-5).

**Tech Stack:** Markdown (meta-prompts, knowledge base), YAML (methodology presets, ADR frontmatter)

**Spec:** `docs/superpowers/specs/2026-03-14-meta-prompt-architecture-design.md`

**Scope:** This plan covers **content authoring only** — meta-prompts, knowledge base, methodology presets, ADRs, and domain model updates. The assembly engine (CLI code that loads and assembles prompts at runtime) and CLI command updates are separate implementation plans that depend on these content artifacts existing first.

---

## File Structure

### New Files (67 total)

**pipeline/** — 32 meta-prompt files