Launching today
RoBrain
Shared AI memory that stops agents from repeating mistakes
19 followers
Shared AI memory that stops agents from repeating mistakes
19 followers
RoBrain is open-source shared memory for teams using AI agents. It captures every decision and the alternatives you ruled out — automatically, across every developer's sessions — and flags when a new decision contradicts an old one. Your team revisits past decisions intentionally, with the original rationale in front of them, instead of re-litigating from zero. Works across Claude Code, Cursor, and Copilot.
RoryPlans
Hey Product Hunt 👋
I built RoBrain because I kept watching my AI agent re-suggest things my team had already ruled out. Every new session, a clean slate — and a clean slate means re-litigating decisions we'd already made.
The fix everyone reaches for is CLAUDE.md (or its equivalent for other agents). It works for a week, then drifts. Nobody updates it. Claude Code's Auto-Memory writes notes automatically now, but they live on your laptop — your teammates' agents have no idea what you decided.
And critically — none of these capture the vetoes. The reasons something was rejected. That's the part that matters most when an agent suggests the same idea three months later.
RoBrain captures decisions as structured data, including what was rejected and why:
json{
"decision": "Use Zustand for state management",
"rationale": "Redux caused re-render issues in cart",
"rejected": [{ "option": "Redux", "reason": "re-render perf issues in cart" }],
"files_affected": ["src/store/cart.ts"]
}
Every developer's sessions feed one shared decision history. Next time your teammate's agent suggests Redux, RoBrain surfaces the prior rejection with the original rationale — so your team revisits intentionally instead of from zero.
@adelina_martiniuc Hi Adelina, Congrats on the launch. How do you differ from openviking and that new palace thing from the actress?
@adelina_martiniuc Also, the manual process of "review the context in *folder"
RoBrain
I jump between Claude Code, Cursor, and Codex all day and the painful part isn’t the models, it’s context. None of your sessions share the same brain, so you end up patching CLAUDE.md and Cursor rules just to repeat yourself.
RoBrain captures more than “what we picked”: it keeps the reasons, the rejected options, and the team context — so when someone else (and their agent, in their stack) takes over, you’re not re-litigating yesterday’s architecture debate from a blank chat.