# Refactoring From Analysis Implementation Plan

> **For agentic workers:** REQUIRED SUB-SKILL: Use superpowers:subagent-driven-development (recommended) or superpowers:executing-plans to implement this plan task-by-task. Steps use checkbox (`- [x]`) syntax for tracking.

**Goal:** Add a durable, type-directed refactoring workspace whose performance items require retained baseline and after evidence.

**Architecture:** `refactor-queue.yaml` orders bounded items, while `/shared/refactor-detail/<project>/<item>/work-item.json` owns detailed item metadata and evidence references. A dedicated verifier enforces type-specific hard gates, especially the performance baseline/after contract.

**Tech Stack:** Markdown/YAML/JSON, Python unittest, existing `/shared` workspace.

**Spec:** `docs/superpowers/specs/2026-08-28-refactoring-from-analysis-design.md`

## Global Constraints

- Application source remains read-only during documentation/refactor planning.
- Performance claims require retained raw baseline and after evidence under equivalent conditions.
- Existing documentation and user files must not be deleted or reset.
- No legacy workspace dependency may be introduced.

---

### Task 1: WorkItem evidence verifier

**Files:**
- Create: `tools/verify_refactor_work_item.py`
- Test: `tools/tests/test_verify_refactor_work_item.py`

- [x] **Step 1: Write failing tests for type validation and performance evidence gates**
- [x] **Step 2: Run tests and observe missing verifier failure**
- [x] **Step 3: Implement the verifier**
- [x] **Step 4: Run verifier tests green**

### Task 2: Refactoring skill and contracts

**Files:**
- Create: `.agents/skills/refactoring-from-analysis/SKILL.md`
- Create: `.agents/skills/refactoring-from-analysis/references/work-item-contract.md`
- Create: `.agents/skills/refactoring-from-analysis/references/type-strategies.md`
- Create: `.agents/skills/refactoring-from-analysis/references/performance-evidence-contract.md`
- Create: `.agents/skills/refactoring-from-analysis/references/evidence-contract.md`

- [x] **Step 1: Encode queue eligibility and one-bounded-item rule**
- [x] **Step 2: Encode type-directed execution and verification**
- [x] **Step 3: Encode performance evidence hard gate**

### Task 3: Durable queue/detail templates

**Files:**
- Create: `/shared/codebase/refactor-queue.yaml`
- Create: `/shared/refactor-detail/README.md`
- Create: `/shared/refactor-detail/_templates/work-item.json`
- Create: `/shared/refactor-detail/_templates/plan.md`
- Create: evidence and verification template directories

- [x] **Step 1: Create queue SSOT**
- [x] **Step 2: Create WorkItem/evidence templates**

### Task 4: Pipeline integration and verification

**Files:**
- Modify: `AGENTS.md`
- Modify: `tools/verify_pipeline.py`
- Modify: `tools/tests/test_verify_pipeline.py`

- [x] **Step 1: Add refactor paths to pipeline verification**
- [x] **Step 2: Add instructions for the new stage**
- [x] **Step 3: Run all verifier and terminal renderer tests**
- [x] **Step 4: Run live workspace verification**