Hey Product Hunt! 👋
Andy here, founder of ByteRover.
Over the past few months, we’ve seen developers try to scale autonomous agents (like OpenClaw and native Ollama setups) and hit a big brick wall: Agent amnesia
An agent solves a problem or writes a script, and then immediately forgets the context. To fix this, teams are putting entire codebases into huge vector databases or blindly spawning massive context windows, resulting in insane API token bills and VRAM crashes.
We are tired of these manual tasks. So we made it Memory skills for OpenClaw.
This is a defined, file-based memory system (.brv/context-tree) that resides directly in your local environment.
How it works:
🧠Select recovery: Instead of blindly injecting everything, ByteRover proactively corrects decisions and feeds the agent. Absolutely What he needs to know.
📉 Token burn cuts: Our users are seeing a ~40-70% reduction in token usage as signals remain noise-free.
📂 Local and Portable: Your repository is version controlled by Git, preventing silent context escalation. What Git did for code, we are doing for the AI ​​context.
We have seen. 26k+ downloads from OpenClaw power users In the past week, Hitting 92.19% retrieval accuracy on the LoCoMo benchmark.
I would love the community’s feedback on our architecture. Leave any questions below and I’ll be here all day to answer them! 👇