QevosAgent is a free, open-source local AI agent supporting Qwen3, Qwen3.6 27B, DeepSeek V4, Gemma 4, Claude and more. Persistent memory, tool calling, web dashboard. One-click install on Windows/macOS/Linux β no WSL needed. Zero API cost, data stays on your machine. A powerful openclaw & hermes agent alternative.
No reviews yetBe the first to leave a review for QevosAgent
Maker
π
Everyone's talking about vibe coding, but what about vibe working? We built QevosAgent to handle more than just code β scientific research, file management, software setup, the whole daily workflow.
The key is smooth human-AI collaboration. That's why QevosAgent has a built-in app system, file manager, and visual dashboard β so the agent and you work together, not against each other.
It runs locally on Windows, macOS, and Linux. No cloud, no WSL, no Docker. Just install and go.
What would you want a local AI agent to do for you?
Report
how does the persistent memory actually work between sessions, is it stored locally as plain files or something more structured
Report
Maker
@ramazaner8kΒ Currently QevosAgent uses a very simple way to do the presistent memory. It has three stage . 1. the memory_macro, which records the memory according the field. eg. the circuit design work, the network manage work, etc. The agent will determine if need to append this at the end of task. All of the memory_macro will be put into the prefill context (2) The task summery. at the end of each task. the agent will write a summary. The last 15 summary will be put into the prefill context. (3) the full context of the task. This will store into disk in realtime and it will not be put into the prefill context, but agent can check it if agent want .
Actually QevosAgent can use more tools to get more complex memory search, you can ask it to evolve the tools to do it.
how does the persistent memory actually work between sessions, is it stored locally as plain files or something more structured
@ramazaner8kΒ Currently QevosAgent uses a very simple way to do the presistent memory. It has three stage . 1. the memory_macro, which records the memory according the field. eg. the circuit design work, the network manage work, etc. The agent will determine if need to append this at the end of task. All of the memory_macro will be put into the prefill context (2) The task summery. at the end of each task. the agent will write a summary. The last 15 summary will be put into the prefill context. (3) the full context of the task. This will store into disk in realtime and it will not be put into the prefill context, but agent can check it if agent want .
Actually QevosAgent can use more tools to get more complex memory search, you can ask it to evolve the tools to do it.