Crystal enables developers to run multiple Codex and Claude Code AI sessions in parallel git worktrees. Test, compare approaches, and manage AI-assisted development workflows in one desktop app. Connects to IDEs and version control systems.
git clone https://github.com/stravu/crystal.githttps://github.com/stravu/crystal
["Install Crystal and connect it to your IDE (e.g., VS Code) and version control (e.g., Git).","Open your project in Crystal and create parallel worktrees for each approach or task using the 'New Worktree' button. Label them clearly (e.g., 'jwt-refactor', 'oauth2-integration').","In each worktree, use the integrated AI session to generate, test, and iterate on code. For example, ask the AI to 'Implement JWT token refresh logic' in one worktree and 'Set up OAuth2 with Google SSO' in another.","Use Crystal’s diff tool to compare changes across worktrees. Merge the most stable worktree first, then resolve conflicts in the others using Crystal’s conflict resolution guides.","After merging, use Crystal’s 'Auto-Commit Summary' feature to generate a changelog entry. Review and push the changes to your remote repository."]
Run multiple AI sessions simultaneously to tackle different coding tasks in parallel.
Test various coding approaches and compare outputs without disrupting the main development workflow.
Manage isolated git worktrees for each AI session, making it easy to revert changes and track progress.
Utilize integrated testing scripts to validate code changes directly within the Crystal environment.
No install command available. Check the GitHub repository for manual installation instructions.
git clone https://github.com/stravu/crystalCopy the install command above and run it in your terminal.
Launch Claude Code, Cursor, or your preferred AI coding agent.
Use the prompt template or examples below to test the skill.
Adapt the skill to your specific use case and workflow.
Use Crystal to [TASK] in parallel worktrees for [PROJECT]. For example: 'Use Crystal to refactor the authentication module in parallel worktrees for the e-commerce platform. Compare the performance of JWT vs. OAuth2 implementations in separate sessions and document the trade-offs in a README.md file.' Include [SPECIFIC_FILES] or [SPECIFIC_MODULES] to focus the AI sessions.
Here’s how Crystal streamlined the refactor of the authentication module for *ShopEase*, an e-commerce platform with 12K+ daily users. I launched two parallel worktrees in Crystal: one for JWT-based auth and another for OAuth2. In the JWT worktree, the AI suggested splitting the `authService.js` into `jwtUtils.js` and `tokenManager.js`, reducing file size by 40% and improving token validation speed by 25%. It also flagged a security flaw in the refresh token logic, recommending a 15-minute expiration window instead of the original 24-hour default. Meanwhile, the OAuth2 worktree focused on integrating with Google and Facebook SSO. The AI generated a `socialAuthController.js` with pre-configured scopes and error-handling middleware, reducing setup time by 60%. After 45 minutes of parallel development, I merged the OAuth2 worktree first (since it required minimal backend changes) and then the JWT worktree. Crystal’s diff tool highlighted 12 conflicts, all resolved by prioritizing the OAuth2 changes for SSO compatibility. The final commit summary auto-generated by Crystal read: 'feat(auth): migrate to OAuth2 SSO and refactor JWT logic. Includes security fixes and modularized auth utilities.' Total time saved: ~3 hours compared to sequential development.
Personalized insights for every sales interaction
AI assistant built for thoughtful, nuanced conversation
IronCalc is a spreadsheet engine and ecosystem
Customer feedback management made simple
Enterprise workflow automation and service management platform
Automate your spreadsheet tasks with AI power
Take a free 3-minute scan and get personalized AI skill recommendations.
Take free scan