braedonsaunders/codeflowPublic

CodeFlow

Visualize Your Codebase Architecture in Seconds

Zero setup. No installation. Just paste a GitHub URL.

Try it Now · Report Bug · Request Feature

Ever opened a new codebase and felt completely lost? CodeFlow turns any GitHub repository or local codebase into an interactive architecture map in seconds.

• No installation required — runs entirely in your browser
• No data collection — your code never leaves your machine
• No accounts — just paste a URL or select local files and go
• Works offline — analyze local files without internet

Paste URL / Select Files -> See Architecture -> Make Better Decisions

See how your files connect at a glance. Click any node to highlight its dependencies. Drag, zoom, and explore. The Code view keeps that map and opens the selected file plus its connected files as full-file cards on the canvas, grouped by directory.

"If I change this file, what breaks?" — CodeFlow answers this instantly. Select any file and see exactly how many files would be affected by changes.

Know who owns what. See the top contributors for any file based on git history. Perfect for code reviews and knowing who to ask.

Automatic detection of:

• Hardcoded secrets & API keys
• SQL injection vulnerabilities
• Dangerous eval() usage
• Debug statements in production code

Test files, fixtures, docs/, and common tooling directories (.github/, .claude/, scripts/) are excluded from the XSS/shell-execution checks, since findings there don't reflect the shipped product's attack surface. The hardcoded-secret check exempts only tests, fixtures, and docs: CI workflows, hooks, and deploy scripts are executable code, so a real credential there is still a real leak.

Automatically identifies:

• Singleton patterns
• Factory patterns
• Observer/Event patterns
• React custom hooks
• Anti-patterns (God Objects, high coupling)

Get an instant A-F grade for your codebase based on:

• Dead code percentage
• Circular dependencies
• Coupling metrics
• Security issues

Color files by commit frequency to see which parts of your codebase are most actively developed.

Paste a PR URL to see exactly which files it affects and calculate the blast radius of proposed changes.

Health grade, scale, fragility, and hidden costs as a self-updating SVG on your README — recomputed every merge, with optional thermal-receipt PR comments. See card/.

Point CodeFlow at an Obsidian vault or any markdown directory to see notes as a connected graph. Both [[wiki-links]] and [text](./relative.md) links become edges; each note is a note-layer node (distinct color) with a dependencies[] array in the JSON export.

Analyze code directly from your computer without uploading to GitHub:

• Privacy First: Your code never leaves your machine
• Offline Support: Works without internet connection
• Drag & Drop: Simply drag files or folders to analyze
• Folder Scanning: Recursively analyze entire project structures
• Exclude Patterns: Skip attachments, caches, generated assets, and other irrelevant paths before scanning
• Instant Results: All processing happens in your browser

A GitHub Action that drops an auto-updating SVG card on your README, recomputed on every merge by the same analyzer as the web app. Five styles, accent presets, opt-in PR receipts, and a privacy mode for public repos. The card adapts to the viewer's light/dark theme automatically.

See card/ for setup, or jump to the style gallery below.

Your code stays on your machine. CodeFlow:

• Runs 100% in the browser
• Makes API calls directly from your browser to GitHub
• Never stores your code or tokens
• Works with private repos (just add your token locally)
• No analytics or tracking

Your GitHub token (if used) is only stored in your browser's memory and is cleared when you close the tab.

Just visit CodeFlow and paste any GitHub URL.

# Clone the repo
git clone https://github.com/braedonsaunders/codeflow.git

# That's it! Just open index.html in your browser
open index.html

No build process. No npm install. Clone the whole repository: index.html loads pinned, checked-in browser dependencies from vendor/, so a fresh local launch works without a network connection.

From this repo (or npx once published): npx codeflow . serves the same index.html UI, watches the folder, and opens it in your browser.

You can now analyze code directly from your local machine without uploading to GitHub:

1. Open CodeFlow in your browser
2. Click the "Open Folder" button
3. Select the folder or files you want to analyze
4. CodeFlow will process them entirely in your browser

Perfect for:

• Private projects you don't want to upload
• Offline development
• Quick local analysis before committing
• Working with sensitive code

Just paste: facebook/react
Or full URL: https://github.com/facebook/react

1. Create a GitHub Personal Access Token with repo scope
2. Paste it in the Token field
3. Analyze your private repos

Click the "Open Folder" button to analyze code from your computer:

• Folder Analysis: Select a folder to analyze all supported files recursively
• File Selection: Choose specific files to analyze
• Drag & Drop: Drag files or folders directly onto the page
• Custom Excludes: Add patterns like uploads/**, **/cache/**, or *.png before scanning

Generated dependencies, caches, build output, local worktrees, and test artifacts are pruned automatically (for example node_modules, .next, .turbo, .local, dist, and playwright-report). Individual files over 2 MB remain visible in results but their contents are not parsed.

All processing happens locally in your browser - nothing is uploaded.

After analysis, click the "Share" button to copy a shareable link. Anyone can re-run the same analysis.

Export your analysis in multiple formats for further processing:

• JSON Report - Complete analysis data including:

  • Repository metadata and health score
  • All files with functions, dependencies, and churn data
  • Complete function statistics with callers and usage metrics
  • Security issues, patterns, and architecture issues
  • Duplicate code detection and layer violations
  • Suggestions and recommendations
  • Language breakdown and folder structure

  Perfect for programmatic analysis, CI/CD integration, or custom reporting tools.

• Markdown Report - Human-readable formatted report

• Plain Text Report - Simple text format

• SVG Image - Export the dependency graph visualization

• PDF Document - Export the dependency graph as a printable PDF

• Raw JSON - Simplified data export

Click the "Export" button in the top bar after analysis to access all export options.

CodeFlow extracts functions and analyzes dependencies for:

Pascal support is heuristic. It recognizes routines and uses unit dependencies while filtering strings and Pascal comment forms; unsupported language features fall back gracefully.

GitHub API has rate limits:

• Without token: 60 requests/hour
• With Personal Access Token: 5,000 requests/hour
• With GitHub App: 5,000 requests/hour per installation

1. Create a GitHub Personal Access Token with repo scope
2. Paste it in the Token field
3. Analyze your private repos

For teams and organizations, GitHub App provides better security and higher rate limits:

1. Create a GitHub App with repository permissions
2. Install the app on your organization or personal account
3. Generate an installation access token
4. Paste the token in the Token field

Benefits of GitHub App:

• Fine-grained permissions control
• Revocable access per installation
• Higher rate limits (5,000 requests/hour)
• Audit logging and security monitoring
• No need to share personal credentials

For larger repositories or team usage, we recommend using GitHub App authentication.

┌─────────────────────────────────────────────────┐
│                   CodeFlow                      │
├─────────────────────────────────────────────────┤
│  ┌──────────┐  ┌──────────┐  ┌──────────┐       │
│  │  Parser  │  │  GitHub  │  │    D3    │       │
│  │  Module  │  │   API    │  │  Graph   │       │
│  └──────────┘  └──────────┘  └──────────┘       │
│        │              │              │          │
│        └──────────────┼──────────────┘          │
│                       │                         │
│              ┌────────▼────────┐                │
│              │   React App     │                │
│              │ (Static Assets) │                │
│              └─────────────────┘                │
└─────────────────────────────────────────────────┘

Zero build dependencies to install. Browser libraries and required Tree-sitter WASM files are pinned under vendor/, with source URLs, SHA-256 digests, and license notices. To refresh them:

node scripts/vendor-browser-deps.mjs

The vendored runtime includes:

• React 18
• D3.js 7
• Babel (for JSX)

We love contributions! Here's how:

1. Fork the repo
2. Make your changes to index.html
3. Test locally (just open in browser)
4. Submit a PR

Node.js unit tests live under tests/ and run with no dependencies:

node --test tests/*.test.mjs tests/*.smoke.js

tests/verify-brain-vault.mjs is an optional end-to-end script that always verifies the bundled fixtures and will also scan a real local vault when you explicitly set BRAIN_VAULT=/path/to/vault.

• Add support for more languages
• Improve function extraction regex
• Add more design pattern detection
• Export to different formats (PNG, PDF)
• Add code complexity metrics

Q: How does it work without a backend?

CodeFlow runs entirely in your browser. It calls the GitHub API directly from your browser and processes everything client-side.

Q: Is my code safe?

Yes. Your code is fetched directly from GitHub to your browser. Nothing is sent to any server we control. The app and its pinned browser dependencies are checked into this repository for inspection.

Q: Can I use it offline?

Yes. Clone the complete repository, open index.html, then use the local file analysis feature. All required browser and parser assets are checked in, so even a cold first load works without internet. GitHub URL analysis still requires access to GitHub's API.

Q: Why is analysis slow?

We make individual API calls for each file to get content. With a token, you get higher rate limits and faster analysis.

Q: How accurate is the dependency analysis?

It's heuristic: unique functions are matched by name, while duplicate function names are scoped by file and explicit imports before edges are created. It may still miss dynamic imports or renamed runtime references, so it's designed for a quick overview, not 100% accuracy.

All examples below are real cards rendered by the CodeFlow Card Action against this very repo. Pick one and drop it on your README.

For public READMEs where you'd rather show data than a letter grade. The card stays informational — files, functions, LOC, languages, tests — without the judgmental bits.

The accent recolors the sparklines, links, and pin. Presets: purple (default), teal, cyan, green, pink, blue, amber, red. Or pass any CSS color (e.g. #ff6b6b).

hero with show-grade: false:

Everything: grade, scale, language breakdown, composition (connections, tests, folders, function stats, patterns), top folders, fragility, hidden costs.

detailed with show-grade: false:

If you find CodeFlow useful, please star the repo.

MIT License — use it however you want.