- Extension points span UI, gateway, server and execution layers across the full runtime lifecycle
- Packages are hot-swappable, versioned and rollback-capable without touching existing app code
- WorkSwarm shows agents generating and installing extension packages on demand
AI CURATED HOT STORY · 11/13/2026
openJiuwen Launches RSI-Native AgentOS with Hot-Swappable Extensions
openJiuwen has launched AgentOS, an operating system designed for recursive self-improvement (RSI) that enables hot-swappable extensions across the entire agent lifecycle. The OS provides extension points at the UI, gateway, server, and agent execution layers, allowing new features to be dynamically plugged in as extension packages without modifying existing application code. It also supports version management and rollback, significantly lowering the barrier for iterative agent functionality.
A demo application called WorkSwarm, built on the OS, showcases an agent autonomously generating and installing system plugins based on natural language requests—such as transforming a workspace into a desktop pet—achieving a self-extension loop of "what you think is what you get." While public materials are demo-centric and lack quantitative benchmarks, the architecture offers a practical engineering paradigm for agents to evolve from passive executors to proactive self-improvers.
Industry-wide, openJiuwen's AgentOS materializes RSI concepts into actionable extension mechanisms, potentially shifting agent development from hardcoded integration to dynamic plugin ecosystems. The hot-swapping and rollback design also enhances safety and maintainability, laying groundwork for standardized extension protocols.
Hourly heat curve
13 fully observed hoursThe line compares only sources observed throughout; gap hours are interpolated to keep the trend continuous.
Coverage timeline
Total 2 reports · Latest first- 量子位官网T2·62 pts
- 量子位 (微信公众号)T2·62 pts
- Extension points span UI, gateway, server and agent execution, with versioned packages and rollback
- WorkSwarm serves as the reference app, showing an agent generating and installing a system plugin on request
- Pitches a self-extension loop, though public material is demo-heavy with little quantitative evaluation