Skip to content
Documentation
Skill

/ghm-template-sync

Detect template version drift and guide migration to the latest template version. Compares current repo against the template, identifies what's outdated, and automates safe updates while protecting product-specific content. Triggers on requests to sync with template, update

From plugin
prd-driven-context-engineering
193100 skills7 agents
Install
$ npx -y skills add mattgierhart/PRD-driven-context-engineering --skill ghm-template-sync --agent claude-code

How it fires

How this skill gets triggered: by you, by Claude, or both.

  • Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.Auto-invocation is when the right skill fires by itself at the right moment, driven by a FLOW.md router and a hook, instead of you invoking it by name. It is the difference between a skill being installed and a skill actually getting used.Read the full definition →
  • You can call itInvoke it directly when you want it.
  • Slash command/ghm-template-sync

Context preview

The summary Claude sees to decide when to auto-load this skill.

Detect template version drift and guide migration to the latest template version. Compares current repo against the template, identifies what's outdated, and automates safe updates while protecting product-specific content. Triggers on requests to sync with template, update

SKILL.md

ghm-template-sync.SKILL.md
name: ghm-template-sync
description: >
  Detect template version drift and guide migration to the latest template version.
  Compares current repo against the template, identifies what's outdated, and automates
  safe updates while protecting product-specific content.
  Triggers on requests to sync with template, update template version, check for template
  drift, or when user asks "sync template", "update to v3", "template drift", "check template version".
disable-model-invocation: true
context: fork
allowed-tools:
  - Read
  - Write
  - Edit
  - Glob
  - Grep
  - Bash

Template Sync

Detect template version drift and migrate to the latest template version.

Workflow

Phase 1: Detect Current State

1. Read `.claude/VERSION` (if it exists) to get current template version 2. If no VERSION file, check `CLAUDE.md` frontmatter for `template_version` 3. If neither exists, assume v1.0.0 (pre-versioning) 4. Report: "Your repo is on template v{X}. Latest is v{Y}."

Phase 2: Diff Against Template

Compare your repo structure against what the current template version expects:

**Check for missing files:**

  • `.claude/VERSION`
  • `.claude/domain-profile.yaml`
  • `.claude/hooks/HOOK_CONTRACT.md`
  • `.claude/hooks/context-validation.sh`
  • `.claude/hooks/context-density-gate.sh`
  • `.claude/hooks/sot-update-trigger.sh`
  • `CHANGELOG.md`
  • `MIGRATION.md`

**Check for stale files (should be removed):**

  • `.claude/hooks/context-validation.py`
  • `.claude/hooks/context-density-gate.py`
  • `.claude/hooks/sot-update-trigger.py`
  • `.claude/agents/HORIZON.md` (replaced by subdirectory)
  • `.claude/agents/STUDIO.md`
  • `.claude/agents/DEVLAB.md`
  • `.claude/agents/METRO.md`

**Check for structure issues:**

  • `settings.json`: Does it use 3-level nesting? Are timeouts in seconds?
  • Agent directories: Do `horizon/`, `studio/`, `devlab/`, `metro/` subdirectories exist with `AGENT.md` + `MEMORY.md`?
  • EPIC template: Does it use semantic headers (not numbered)?
  • Frontmatter: Do key files have `template_version`?

Phase 3: Generate Migration Plan

Output a table:

| File | Status | Action | Risk | |------|--------|--------|------| | `.claude/VERSION` | Missing | Copy from template | None | | `.claude/hooks/*.py` | Stale | Delete (replaced by .sh) | None | | `.claude/agents/HORIZON.md` | Stale | Split into horizon/AGENT.md + MEMORY.md | **Preserve MEMORY.md content** | | `settings.json` | Outdated | Update hook nesting + commands | Check for custom hooks | | `CLAUDE.md` | Missing frontmatter | Add `template_version: "3.0.0"` | None |

Phase 4: Execute Safe Updates

**Auto-safe** (do without asking):

  • Create `.claude/VERSION` with current template version
  • Add `template_version` frontmatter to files that lack it
  • Report what was done

**Confirm first** (show diff, ask user):

  • Update `settings.json` hook configuration
  • Restructure agent files (must preserve MEMORY.md)
  • Delete stale Python hooks
  • Update EPIC template headers

**Never touch** (product-specific):

  • `PRD.md` content (only add frontmatter)
  • `SoT/*.md` content
  • `epics/EPIC-*.md` content (only update headers on closed EPICs)
  • `.claude/agents/*/MEMORY.md` content
  • `README.md` content

Phase 5: Verify

After all changes: 1. Test all shell hooks produce valid JSON 2. Verify no Python hooks remain 3. Confirm `settings.json` uses correct nesting 4. Check agent subdirectories have both AGENT.md and MEMORY.md 5. Report summary of changes made

Version Migration Matrix

| From | To | Key Changes | |------|----|-------------| | v1.0.0 | v2.0.0 | Add skills, hooks, agents, SoT standardization | | v2.0.0 | v3.0.0 | Shell hooks, agent subdirs, semantic EPICs, versioning, domain-profile | | v1.0.0 | v3.0.0 | All of the above (cumulative) |

Safety Rules

1. **NEVER overwrite MEMORY.md** -- these contain product-specific agent memory 2. **NEVER modify SoT content** -- only add section markers or frontmatter 3. **NEVER modify PRD.md content** -- only add frontmatter version 4. **ALWAYS show diff before destructive changes** (deletes, restructures) 5. **ALWAYS verify hooks work** after updating settings.json 6. **Commit changes incrementally** -- one commit per phase, not one giant commit

Read more
Ships withprd-driven-context-engineering

PRD-driven Context Engineering: A systematic approach to building AI-powered products using progressive documentation and context-aware development workflows

Get the whole plugin