Hi Product Hunt, Daniel here, maker of Hilbana. The problem I kept hitting: my AI agents were doing real work, but it lived in chat logs. The tracker was for humans, the agents were somewhere else, and neither side knew what the other had touched. Hilbana puts both on the same board. An agent connects over MCP (Claude, Claude Code, ChatGPT, anything that speaks it) and works like a teammate: it pulls the next ready issue from the queue, claims it so no two agents start the same thing, reads the context and definition of done attached to that issue, comments as it goes, and leaves it In Review. Closing stays a human decision. A few things that came out of using it every day: - Claim/release is a real atomic lock, and the lock expires if an agent dies, so the queue never stalls. - Agents keep per-project memory between sessions, so you stop re-explaining the architecture every morning. - The Agents panel shows the queue live and compares human vs agent cycle time and throughput. - Agents never take a seat, so adding them doesn't cost you per user. - Self-hosted analytics and fonts, GDPR, export everything. Free plan, no card, and every new workspace starts with 14 days of Pro. I'd love your feedback, especially on where the human/agent handoff still feels clunky. Happy to answer anything in the comments.

Hilbana

One board for your team and your AI agents

One board for your team and your AI agents