Install and sync your AI development configurations across Claude Code, Cursor, and OpenAI Codex with a single command.
Define your commands, agents, and templates once β deploy everywhere.
- π Multi-tool Sync β Claude Code, Cursor, and Codex support built-in
- π― Interactive CLI β Beautiful TUI for selecting tools and components
- π¦ Template System β Customizable commands, agents, and doc templates
- πΎ Auto Backup β Never lose your configurations
- π Watch Mode β Real-time sync on file changes
- π οΈ Path Placeholders β Tool-agnostic paths that work everywhere
Run instantly (no installation):
npx mindkit installInstall globally:
npm install -g mindkitVia Homebrew:
brew tap viniciuscarvalho/mindkit
brew install mindkit# Interactive mode - select tools and components
mindkit install
# Quick install to specific tools
mindkit install --tools claude,cursor
# Preview what would be installed
mindkit install --dry-runmindkit init# One-time sync
mindkit sync --source claude --target cursor
# Watch mode - auto-sync on changes
mindkit sync --watchmindkit backup create # Create backup
mindkit backup list # List backups
mindkit backup restore # Restore from backupmindkit list # All components
mindkit list commands # Commands only
mindkit list agents # Agents only
mindkit list tools # Detected toolsmindkit automatically translates configurations for each tool:
| Concept | Claude Code | Cursor | Codex |
|---|---|---|---|
| Commands | ~/.claude/commands/*.md |
.cursor/rules/*.mdc |
~/.codex/AGENTS.md |
| Agents | ~/.claude/agents/*.md |
Embedded in rules | Merged in AGENTS.md |
| Project config | CLAUDE.md |
.cursorrules |
AGENTS.md |
| Name | Description |
|---|---|
create-prd |
Generate Product Requirements Documents |
generate-spec |
Create technical specifications from PRDs |
generate-tasks |
Break down specs into implementable tasks |
| Name | Description |
|---|---|
architect |
Software architecture specialist for system design |
backend-developer |
Backend engineer for APIs and microservices |
code-reviewer |
Expert code review for quality, security, and maintainability |
frontend-developer |
Modern frontend specialist with React/Vue expertise |
mobile-developer |
Cross-platform mobile development specialist |
swift-expert |
Senior Swift developer with SwiftUI expertise |
swift-reviewer |
Code reviewer for Swift/SwiftUI code quality |
swiftui-specialist |
SwiftUI expert for complex UI implementation |
typescript-pro |
TypeScript expert for full-stack development |
ui-designer |
Visual designer for interfaces and design systems |
Templates use placeholders that resolve correctly for each tool:
Use template from {{DOCS}}/specs/prd-template.md
Output to {{PROJECT}}/docs/tasks/prd-{{feature}}/prd.md| Placeholder | Description |
|---|---|
{{DOCS}} |
Documentation directory |
{{PROJECT}} |
Project root |
{{HOME}} |
User home directory |
{{CONFIG}} |
Tool's global config |
Add your own templates in ~/.mindkit/templates/:
~/.mindkit/
βββ templates/
β βββ commands/
β β βββ my-command.md
β βββ agents/
β βββ my-agent.md
βββ registry.yaml
# ~/.mindkit/registry.yaml
version: 1
templates:
- name: my-command
source: commands/my-command.md
type: commands
targets:
claude:
path: ~/.claude/commands/my-command.md
cursor:
path: .cursor/rules/my-command.mdcimport { getAdapter, detectInstalledTools, getAllTemplates } from 'mindkit';
// Detect installed tools
const tools = await detectInstalledTools();
// Map { 'claude' => true, 'cursor' => true, 'codex' => false }
// Get adapter and install
const claude = getAdapter('claude');
const templates = await getAllTemplates();
await claude.install(templates.commands[0], content);See the CHANGELOG for a complete history of changes.
| Version | Date | Highlights |
|---|---|---|
| 1.1.0 | 2026-02-04 | New agents and enhanced templates |
| 1.0.2 | 2026-01-22 | Fix version detection and package.json handling |
| 1.0.1 | 2026-01-20 | Improved tool detection via PATH |
| 1.0.0 | 2026-01-19 | Initial release |
git clone https://github.com/Viniciuscarvalho/mindkit.git
cd mindkit
npm install
npm run build
npm link # Link for local testingMIT Β© Vinicius Carvalho