Community Contribution: Landing Page + Documentation Site

[tody-agent]

Hey @jamesgrugett @brandonchen — 👋

I'm a developer who's been exploring Codebuff and was genuinely impressed by the multi-agent architecture and the 61% pass rate on SWE benchmarks. I decided to build something for the project out of pure curiosity and passion.

What I Built

A complete landing page + documentation site using Astro Starlight:

🌐 Live preview: https://tody-agent.github.io/codebuff-docs/
📦 Source code: https://github.com/tody-agent/codebuff-docs

Landing Page (8 sections)

• Hero with data-driven hook (61% vs 53% benchmark comparison)
• Problem/solution narrative
• Multi-agent architecture diagram
• Feature bento grid (OpenRouter, Custom Agents, SDK, Agent Store, 33 tools)
• 3-step quick start guide
• Performance metrics showcase
• Documentation links
• CTA footer

Documentation Site

• System Architecture — multi-agent system, tech stack, ADRs, security model
• User Guides — CLI, custom agents, SDK integration (step-by-step SOPs)
• API Reference — 33 tools, SDK client API, CLI commands, error codes
• Database Schema — 18 PostgreSQL tables, Drizzle ORM, migration strategy
• Full-text search via Pagefind

Tech Stack

• Astro + Starlight (docs framework)
• Dark theme, responsive, JetBrains Mono + IBM Plex Sans
• GitHub Actions CI/CD for GitHub Pages
• SEO-optimized, accessible

My Ask

I built this entirely for free — no payment needed. The only thing I'd appreciate is keeping the small footer credit to the two free AI skills I used to build this:

"Built with UX Master & DocKit Master — free AI skills by TodyAI"

How to Use

You're welcome to:

1. Fork the repo and deploy it as your official docs site
2. Cherry-pick sections you like and integrate into your existing site
3. Use it as a starting point and customize from there

I'm happy to transfer the repo to the CodebuffAI org if that's easier.

Let me know what you think! Would love to contribute more to the project. 🚀

[VaishnavGunjari]

One potential improvement for the documentation could be an Agent Flow Visualizer that visually shows how Codebuff processes a user prompt through its multi-agent system.

For example, a simple interactive diagram could illustrate:

User Prompt → Planner Agent → Tool Selection → Executor Agent → Final Result

This could help new developers quickly understand the workflow and architecture of the system. Tools like React Flow, Mermaid, or D3.js could be used to implement this in the docs.

If this sounds useful, I'd be happy to help design or implement it.