OpenThunder: AI Codebase Health Check
Own the code AI wrote for you.
OpenThunder runs AI Codebase Health Checks from VS Code, explains architecture, finds risks, and turns cleanup work into verified missions.
From vibe-coded to verified.
What it does
You built an app quickly with Claude, Cursor, Copilot, or another AI tool. Now you need to understand it, trust it, and make it production-ready.
OpenThunder scans your repo, gives it a grade, identifies the biggest risks, and helps you fix them with verified cleanup missions.
The loop:
Open repo → Run Health Check → Get score → Review risks → Create cleanup mission → Verify → Share
Key commands
| Command |
Description |
OpenThunder: Run Health Check |
Scan the current workspace and get a grade + risk report |
OpenThunder: Open Full Report in Dashboard |
View the detailed report in the OpenThunder dashboard |
OpenThunder: Copy Health Summary |
Copy a shareable summary for LinkedIn, X, or your README |
OpenThunder: Explain This Repo |
Open the Architecture Notebook for this repo |
OpenThunder: Create Mission from Selection |
Turn selected code into a verified cleanup mission |
OpenThunder: Open Dashboard |
Open the full OpenThunder dashboard |
OpenThunder: Toggle Inline Completions |
Enable or disable ghost-text completions |
OpenThunder: Copy AI Context Pack |
Copy paste-ready repo context for Claude Code, Cursor, Copilot, or Codex |
OpenThunder: Review Current Diff |
Offline review of your uncommitted changes |
OpenThunder: Generate PR Summary |
PR-ready summary, opened and copied |
OpenThunder: Open Repo in OpenThunder Desktop |
Hand the repo to the OpenThunder Desktop app |
OpenThunder: Install MCP Server (agent knowledge graph) |
One-click install of the OpenThunder MCP server for Claude Code, Cursor, or Codex |
The Health Check panel in the OpenThunder sidebar shows:
- Connection status to the local server
- Current workspace name
- Latest health grade (A to F)
- Score out of 100
- Top 3 risks with severity
- Buttons: Run Health Check, Open Full Report, Copy Share Summary
- One-click mission creation from any risk
Health Check report
Each health check produces a report with:
- Overall grade (A+ to F)
- Category scores: architecture, verification, tests, security, maintainability, documentation, deployment, production readiness, AI code risk
- Architecture summary
- Architecture layers
- Top risks with severity (critical, high, medium, low)
- Recommended cleanup missions
- Production readiness summary
- Study path (which files to read first)
- Shareable summary
Requirements
OpenThunder requires a local server running at http://localhost:7700.
Options:
- Download the desktop app from openthunder.dev; the server starts automatically
- Clone the repo and run
pnpm dev from the project root
Getting started
- Install the extension
- Start the OpenThunder server (
pnpm dev or desktop app)
- Open any repo in VS Code
- Click the OpenThunder icon in the activity bar
- Click Run Health Check
- See your grade and top risks in under 60 seconds
Configuration
| Setting |
Default |
Description |
openthunder.serverUrl |
http://localhost:7700 |
Local OpenThunder server URL |
openthunder.completionsEnabled |
true |
Inline AI completions |
openthunder.provider |
mock |
AI provider for completions: anthropic, groq, openrouter, ollama, mock |
Example share summary
After a health check, copy and share:
My AI-built app scored C+ on OpenThunder.
Architecture: B
Tests: D
Security: C
Production readiness: C+
Top issue: No reliable verification before shipping.
AI helped me build the app. OpenThunder helped me own it.
Privacy
Your code never leaves your machine. OpenThunder runs fully locally. AI calls go directly from your machine to your chosen provider.
Secrets and .env values are never sent to AI providers. The health check is always read-only.
Links
