cc-crossbeam  by mikeOnBreeze

AI assistant for California ADU permit processing

Created 6 months ago
282 stars

Top 92.3% on SourcePulse

GitHubView on GitHub
Project Summary

CrossBeam addresses the critical problem of high rejection rates for California Accessory Dwelling Unit (ADU) permits, which cause significant delays and costs for homeowners and contractors. It acts as an AI-powered assistant, leveraging Claude Opus 4.6 to interpret architectural plans and city corrections letters, generating professional response packages to streamline the resubmission process. The primary benefit is reducing permit turnaround times and associated expenses.

How It Works

CrossBeam employs a "skills-first" design, where an AI agent's domain knowledge is encapsulated in structured reference files (skills) that cover California ADU law, city-specific codes, and workflow guidance. The system architecture separates long-running AI agent tasks from frontend requests using a Cloud Run orchestrator that launches isolated, ephemeral environments on Vercel Sandboxes. This setup allows the agent to interact with documents via vision and perform live web research, while Supabase handles database, storage, and real-time updates to the frontend.

Quick Start & Requirements

  • Primary install/run command: Navigate to frontend and server directories, run npm install, then npm run dev in each.
  • Prerequisites: Node.js 20+, a Supabase project, an Anthropic API key (for Claude Opus 4.6), and a Vercel account.
  • Links: No direct quick-start or demo links provided in the README.

Highlighted Details

  • AI-powered ADU permit assistant built with Claude Opus 4.6.
  • "Skills-first" design for modular and extensible domain knowledge.
  • Architecture utilizes Cloud Run and Vercel Sandboxes to manage long-running AI agent tasks.
  • The entire project was developed using Claude Code.

Maintenance & Community

  • Contact information provided via X/Twitter (@breezwoods) and LinkedIn (Michael Brown).
  • No explicit community channels (e.g., Discord, Slack) or roadmap links are mentioned.

Licensing & Compatibility

  • License: MIT.
  • Compatibility: The MIT license is permissive and generally compatible with commercial use and closed-source linking.

Limitations & Caveats

The "City Pre-Screening" feature is outlined as a future roadmap item and is not yet implemented. The system relies on external services like Supabase, Vercel, and Anthropic's API, requiring separate accounts and potential costs.

Health Check
Last Commit

5 months ago

Responsiveness

Inactive

Pull Requests (30d)
0
Issues (30d)
0
Star History
0 stars in the last 30 days

Explore Similar Projects

Feedback? Help us improve.