AgentBoard
AgentBoard — Kanban for AI agents
Start new thread
Rusl St

AgentBoard - AgentBoard — Kanban for AI agents

by•
AgentBoard is an acceptance-driven development board where AI agents work through a structured pipeline — and humans stay in control. How it works 1. Create a task — describe what you need built 2. AI agent auto-plans — generates an implementation plan and submits for review 3. You approve or reject — with comments that guide the agent 4. Agent implements — writes decision logs explaining key choices 5. You accept or send back — final quality gate before marking done

Add a comment

Replies

Best
Rusl St
Maker
📌
Hey Product Hunt! I built AgentBoard because I was tired of copy-pasting context between Claude Code sessions and reviewing AI-generated PRs with zero audit trail. The core idea: AI agents are productive but unreliable. They need guardrails — not micromanagement. AgentBoard gives agents a clear pipeline (plan → approve → implement → accept) while keeping humans in the loop at every decision point. The secret sauce is decision logs. When an agent completes a task, it documents what it decided and why. Those decisions automatically flow into dependent tasks, so the next agent starts with full context. This is how you get coherent multi-task projects instead of isolated code snippets. It's fully open source and self-hosted. Try the live demo at agentboard.online — one click, no signup. Would love your feedback. What would make this useful for your AI agent workflow?
Avinash Dalvi

@rusl_st This looks great. I’ve been using tools like Claude Code and Cursor, but keeping context across multi-task projects is always the hardest part. The way you flow decisions into dependent tasks is exactly what’s missing right now.

I love that it's open source and self-hosted. Quick question: How does it handle context limits when projects get large and the decision logs start piling up? Would love to try this on my next project.

Mohammed Fazalur Rahman

“This is a smart approach to human-in-the-loop AI development. The approval checkpoints and decision logs add real accountability