Tutorial: Migrate Existing Configs¶
Time: 15 minutes | Level: Intermediate
Convert your existing AI tool configurations into a shareable DevSync package.
What You Will Learn¶
- Audit existing AI tool configurations
- Extract them into a DevSync package
- Upgrade v1 packages to v2 format
- Distribute to your team
Prerequisites¶
- DevSync installed (
pip install devsync) - LLM configured (
devsync setup) - An existing project with AI rules, MCP configs, or other tool-specific files
Step 1: Audit Existing Configs¶
Check what AI configurations your project already has:
Then look at the tool-specific directories:
# Claude Code
ls .claude/rules/ 2>/dev/null
# Cursor
ls .cursor/rules/ 2>/dev/null
# Windsurf
ls .windsurf/rules/ 2>/dev/null
# GitHub Copilot
ls .github/instructions/ 2>/dev/null
# Single-file configs
ls AGENTS.md CONVENTIONS.md GEMINI.md 2>/dev/null
Step 2: Extract into a Package¶
DevSync reads all detected configs and produces a package:
Extracting practices from /home/user/my-project...
Scanning: .claude/rules/ (4 files)
Scanning: .cursor/rules/ (3 files)
Scanning: MCP configurations (2 servers)
Extracted 5 practice declarations
Extracted 2 MCP servers
Package written to: ./migrated-standards/devsync-package.yaml
Step 3: Review the Package¶
Check the extracted practices:
Verify that:
- Practice names are clear and descriptive
- Principles accurately capture your standards
- MCP server credentials are properly declared (not hardcoded)
Edit the manifest if needed to refine the extracted content.
Step 4: Test the Package¶
Install into a clean project to verify:
mkdir /tmp/test-migration && cd /tmp/test-migration
git init
devsync install ~/my-project/migrated-standards
devsync list
Step 5: Distribute¶
Push the package to Git for your team:
cd migrated-standards
git init && git add . && git commit -m "Initial migration"
git remote add origin https://github.com/company/team-standards
git push -u origin main
Upgrading v1 Packages¶
If you have an existing v1 package (ai-config-kit-package.yaml), upgrade it to v2:
This converts v1 instructions into v2 practice declarations while preserving MCP server configurations.
Before (v1)¶
# ai-config-kit-package.yaml
name: old-package
version: 1.0.0
components:
instructions:
- name: python-style
file: instructions/python-style.md
description: Python style guide
After (v2)¶
# devsync-package.yaml
name: my-standards
version: 1.0.0
practices:
- name: python-style
intent: Consistent Python code formatting and naming
principles:
- Line length 120 characters
- Use black for formatting
- snake_case for functions
tags: [python, style]
File-Copy Mode¶
If you don't have an LLM configured, use file-copy mode:
# Extract files verbatim
devsync extract --output ./pkg --name my-pkg --no-ai
# Install files verbatim
devsync install ./pkg --no-ai
This copies files directly without AI extraction or adaptation.
Next Steps¶
- Team Config Repository -- host and maintain your standards
- Custom Packages -- add hooks, commands, and resources
- Package Examples -- see complete package examples