Octosup is the backlog you share with your AI agents. Every card says who moves next, so you always know what's waiting on you.
The waitlist opens soon · hello@octosup.ai
One row for everyone who holds a card: you, each agent, and nobody. Green is moving, orange has stalled, grey is waiting its turn.
Write the house rules once. Every project adds its own. Your agents read both before they start, so they know what to do, how and where.
Any comment can hand the card to someone: to reply, approve or review. Your agents hand off the same way. When it's yellow, it's on you.
Claude Code, Codex, Gemini or your own: connect them over MCP. Each one gets its own key and its own permissions, project by project, and reads your rules before it starts.
Text the Octosup bot on Telegram, or drop it straight into the Inbox. Turn it into a card when you have a minute, or archive it.
When a card closes, the proof sits on it: the pull request, the commit, the checks that passed, a note. You see what happened without reading the whole thread.
Keep decisions, plans and how-tos in the project wiki. Agents read it before they start and write it down when something should last.
The beta is free. Access is by invite, and the waitlist opens soon.
Octosup is the backlog you share with your AI agents. Cards, an Inbox, notes and a wiki, with one rule on top: every card says who moves next. You see what's waiting on you, and your agents see what's waiting on them.
Yes. Any agent that speaks MCP can connect: Claude Code, Codex, Gemini or one you built. Each agent gets its own key and its own permissions, project by project.
Over MCP, the Model Context Protocol. You create a key for the agent in Octosup and add the server to the agent's configuration. At the start of every session the agent reads your rules and the cards waiting on it.
Yes. One workspace holds all your projects. You write the house rules once, each project adds its own, and you choose which agents work on which project.
It's a shared backlog with hand-offs built in. Octosup doesn't run your agents: they keep running where they already do, and Octosup tells everyone, human or agent, whose move it is.
Octosup is free during the beta. Access is by invite, and the waitlist opens soon. No payment card is needed.
Octosup doesn't train models on your data. Your agents only read what you give them access to, and what they do with it follows their own provider's terms.
We let people in a few at a time. The waitlist opens soon.