RefactorSAFE

Claude Code plugin that generates individualized knowledge systems from conversation. You describe how you think and work, have a conversation and get a complete second brain as markdown files you own.

The instruction file, verbatim from the audited commit — this is the text the model reads, and the surface the audit's instruction layer examines. Quoted here so you can judge it without cloning anything.

---
name: refactor
description: Plan vault restructuring from config changes. Compares config.yaml against derivation.md, identifies dimension shifts, shows restructuring plan, executes on approval. Triggers on "/refactor", "restructure vault".
version: "1.0"
generated_from: "arscontexta-v1.6"
user-invocable: true
context: fork
allowed-tools: Read, Write, Edit, Grep, Glob, Bash
argument-hint: "[dimension|--dry-run] — focus on specific dimension or preview without approval prompt"
---

## Runtime Configuration (Step 0 — before any processing)

Read these files to configure domain-specific behavior:

1. **`ops/derivation-manifest.md`** — vocabulary mapping, dimension positions, platform hints
   - Use `vocabulary.notes` for the notes folder name
   - Use `vocabulary.note` / `vocabulary.note_plural` for note type references
   - Use `vocabulary.topic_map` / `vocabulary.topic_map_plural` for MOC references
   - Use `vocabulary.inbox` for the inbox folder name

2. **`ops/config.yaml`** — current live configuration (the "after" state)

3. **`ops/derivation.md`** — original derivation record (the "before" state)

If these files don't exist, report error: "Cannot refactor without both ops/config.yaml and ops/derivation.md. These files establish the baseline and current configuration."

---

## EXECUTE NOW

**INVARIANT: /refactor never executes without approval.** The plan is always shown first.

**Target: $ARGUMENTS**

Parse immediately:
- If target is empty: detect ALL config changes and plan restructuring
- If target names a specific dimension (e.g., "granularity", "organization"): focus on that dimension only
- If target is `--dry-run`: show plan without asking for approval

**Execute these phases:**

1. Detect changes between config.yaml and derivation.md
2. Plan restructuring for each changed dimension
3. Show the plan with affected artifacts and risk assessment
4. Execute on approval (unless --dry-run)
5. Validate post-restructuring

**START NOW.** Reference below defines the workflow.

---

## Philosophy

**Configuration changes cascade.**

Changing a dimension is not just editing a value in config.yaml. Each dimension affects multiple artifacts — skills, templates, context file sections, hooks, MOC structure. Changing organization from "flat" to "hierarchical" means restructuring the notes directory, updating MOC templates, adjusting navigation references in the context file, and regenerating skills that reference folder structure.

/refactor makes these cascades visible and manages them. Without it, dimension changes create drift: config says one thing, artifacts say another. With it, every change is planned, approved, and validated.

**The relationship to /architect:** /architect RECOMMENDS changes. /refactor IMPLEMENTS them. /architect analyzes health and friction to propose dimension shifts. When those proposals are approved, /refactor ensures every affected artifact is updated consistently.

---

## Phase 1: Detect Changes

Compare each dimension in `ops/config.yaml` against the same dimension in `ops/derivation.md`:

**Step 1: Read both files**

Read `ops/config.yaml` and `ops/derivation.md` fully. Extract the position for each of the 8 dimensions from both.

**Step 2: Build comparison table**

| Dimension | Derivation Value | Config Value | Changed? | Drift Type |
|-----------|-----------------|--------------|----------|------------|
| granularity | [val] | [val] | [yes/no] | [aligned/misaligned] |
| organization | [val] | [val] | [yes/no] | [aligned/misaligned] |
| linking | [val] | [val] | [yes/no] | [aligned/misaligned] |
| processing | [val] | [val] | [yes/no] | [aligned/misaligned] |
| navigation | [val] | [val] | [yes/no] | [aligned/misaligned] |
| maintenance | [val] | [val] | [yes/no] | [aligned/misaligned] |
| schema | [val] | [val] | [yes/no] | [aligned/misaligned] |
| automation | [val] | [val] | [yes/no] | [aligned/misaligned] |

**Step 3: Check feature flags**

Also compare feature flags between derivation and config:

| Feature | Derivation | Config | Changed? |
|---------|-----------|--------|----------|
| semantic_search | [on/off] | [on/off] | [yes/no] |
| processing_pipeline | [on/off] | [on/off] | [yes/no] |
| self_space | [on/off] | [on/off] | [yes/no] |
| session_capture | [on/off] | [on/off] | [yes/no] |
| parallel_workers | [on/off] | [on/off] | [yes/no] |

**Step 4: Early exit if no changes**

If no changes detected:
```
--=={ refactor }==--

  No configuration drift detected between config.yaml and derivation.md.
  All dimensions and features match the original derivation.

  If you want to explore changes, run /architect for recommendations
  or edit ops/config.yaml directly, then run /refactor again.
```

---

## Phase 2: Plan Restructuring

For each changed dimension, determine ALL affected artifacts. This is the cascade analysis.

### Dimension-to-Artifact Mapping

| Change | Affected Artifacts | What Changes |
|--------|-------------------|-------------|
| **Granularity shift** | Note templates, extraction depth in /reduce, processing skills, context file "Note Design" section | Template body length guidance, extraction granularity settings, composability test thresholds |
| **Organization shift** | Folder structure, MOC hierarchy, context file "Folder Architecture" section, hub MOC | Directory layout, MOC tier count, navigation references |
| **Linking shift** | Semantic search config, /reflect connection density expectations, context file "Connection Finding" section | Search tool availability, link threshold values, discovery layer instructions |
| **Processing shift** | /reduce depth settings, /reflect pass count, pipeline skills, context file "Processing Pipeline" section, config.yaml processing.depth | Extraction thoroughness, connection evaluation depth, chaining mode |
| **Navigation shift** | MOC tier structure, hub MOC, context file "MOC" section, note Topics footers | Number of MOC tiers, hub content, navigation instructions |
| **Maintenan

SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.

What the source does

Filesystem

none-observed

Network

none-observed

Shell

none-observed

Dependencies

pinned

Secrets in source

none-found

Findings (0)

No findings outside the package's declared scope.

Gates applied: no_behavioural_pass.