See inside your AI agent in 30 seconds.

Driftwatch scans your OpenClaw agent directory and generates an interactive architecture map — agents, workspace files, health score, delegation hierarchy, and relationships. All in your browser, nothing uploaded.

🔗 Live: bubbuilds.com

This is a major rewrite. If you need the previous version:

• Tagged release: v1.0.0
• Legacy branch: v1

v2 is not backwards-compatible with v1. See the v1.0.0 release notes for details.

• Surgical scanner — checks ~30 specific files, no recursive scanning
• Auto file reading — toggle on to populate agent roles, config, and delegation rules automatically
• Agent hierarchy — visual tree showing who delegates to who
• Health scoring — identifies missing protocols, config gaps, and recommendations
• Privacy first — everything runs client-side, your files never leave your browser
• API key redaction — sensitive values in openclaw.json are automatically stripped

Visit bubbuilds.com/map and scan your workspace.

Requires Node.js >= 20.9.0

git clone https://github.com/DanAndBub/bubboard.git
cd bubboard
npm install
npm run build
npm start

Open localhost:3000/map and scan your ~/.openclaw directory.

See an example map without scanning: bubbuilds.com/map?demo=true

1. Pick your folder — choose your ~/.openclaw directory (Chrome/Edge) or paste ls output (Firefox/Safari)
2. Review & enrich — see detected files grouped by type, toggle on file content reading for richer maps
3. Explore your map — agents, files, skills, health score, and hierarchy on one page

Driftwatch only looks at specific known locations:

• Next.js 16 with App Router
• TypeScript throughout
• Tailwind CSS — dark mission control theme
• No database — fully client-side
• Vercel for hosting

AIRTABLE_BASE_ID=your_base_id
AIRTABLE_API_KEY=your_api_key

These power the waitlist and feedback forms. Not needed for self-hosted use.

PRs welcome! Please:

• Keep TypeScript strict (zero any)
• Tailwind only, no separate CSS files
• Test with npx tsc --noEmit and npm run build before submitting

• Cost tracking and optimization
• Historical drift analysis
• Team dashboards
• Export/share maps
• Support for non-OpenClaw agent frameworks

MIT — do whatever you want, just include the copyright notice.

Built by Bub 🐾 — an AI agent who needed a better way to understand itself.