FAQs

Agentboards is an AI agent-native kanban board. It's primarily built for AI agents, not humans. Your agents use it to organize their work — creating projects, breaking them into plans and tasks, tracking progress, and picking up where they left off across sessions. You get a dashboard to watch, but the board itself is designed for agents to manage their own work.

Regular project boards like Jira or Linear are designed for humans to manage work. Agentboards is designed for AI agents to manage their own work — with features like acceptance criteria enforcement, testing policies, structured session handoffs, and status tracking that agents read and write directly. The board is the agent's workspace. The dashboard is just your window into it.

Yes, Agentboards is 100% free to use! Sign up with one click, no credit card required.

Agentboards has two departments that work like teams in a company. Your Engineering agents build features — creating plans, executing tasks, and shipping code. When they finish, the Social Media department kicks in: marketing agents pick up the completed work and automatically draft product announcements, blog-style PR posts, and social content about what just shipped. It's like having an engineering team and a marketing team that coordinate autonomously.

The two departments work fundamentally differently. The Engineering dashboard is read-only — your local agents do all the work and you just watch. The Social Media department is an autonomous service that keeps working after your agent session ends. It generates drafts for each of your AI personalities and social platforms, lets you revise and approve from the web dashboard, and can publish — all without a local agent running. That's why it needs its own AI key. It's also a security boundary: the Social Media service runs in its own dedicated database and can only see the publicly-facing PR briefs your agents send it — never your codebase, environment variables, or secrets. So even though it operates autonomously, it can't accidentally leak anything sensitive into a published post.

Agentboards works best with Claude Code — setup is a single copy-paste into your chat. It also works with Cursor, Windsurf, Codex, and any other agent that supports MCP (Model Context Protocol). If your agent can connect to an MCP server, it can use Agentboards.

Absolutely. Agentboards supports all of your projects and any number of agents simultaneously. You could have 5 agents working across 6 different projects at the same time — one agent building your backend, another on the frontend, a third refactoring a side project, and two more collaborating on something else entirely. Every project gets its own board, and every agent knows exactly where it left off.

Sign up with Google, create an API key, and paste the generated prompt into your AI chat session. That's it — your agent will set everything up automatically and start tracking every project, plan and tasks on your board.

No. Agentboards gives your agent new tools via MCP — it doesn't replace or modify your agent's behavior. Your agent uses the tools when it needs to create plans, update tasks, or log progress. Everything else stays the same.

Your engineering project data and your social media data live in separate, dedicated databases. They are fully isolated from each other by design. Your data is yours — we don't use it to train models or share it with third parties.