DeepSeek opensources Super-agent architecture: AI gains autonomous bug-fixing and cross-platform collaboration
Current overview (not historical)
DeepSeek has open-sourced its Super-agent architecture, an AI system designed with continuous, efficient, active, and persistent capabilities. Built on a dual-layer Router-Watcher framework, it supports cross-platform task context retention and subscription mechanisms. In simulated multi-task environments, the median task completion time was 12.4 minutes and the 95th percentile was 15.7 minutes, demonstrating the agent's ability to autonomously discover bugs and submit pull requests. The architecture aims to enable AI agents to operate with greater autonomy across diverse platforms, streamlining workflows that involve code debugging and collaborative tasks.
Written with AI from the reporting · updated 11/2/2026
Expand the detailed fact route and sources
Story route
A station-by-station route from the first public evidence onward; every station carries its own evidence and uncertainty.
- 1Unclassified fact stationOct 31, 00:00recorded Nov 2, 12:55single sourcerevised
DeepSeek开源Super-agent底座架构
Read original source ↗ QbitAI · AI Industry & ResearchReplay this moment in the explorerEvidence & revisions · 1 source records
- Subject:
- deepseek
- Action:
- 开源架构思路与原型代码
- Object:
- Super-agent系统
- Occurred (fact side):
- Oct 31, 00:00
- Published (valid side):
- Oct 31, 21:49
- First seen (known side):
- Nov 2, 12:55
Fact stationSingle source · one-sided claimRevisedCopy may have been revised later; time replay is bounded. Layer labels group reading only, not causation between layers.
Uncertainty marks flag where to check the sources; they are not proven causation.
Report timeline
Explore every side through dated source reports.
- QbitAI · AI Industry & ResearchRelated fact: single sourceSource report revisedDeepSeek open-sources Super-agent prototype with multi-session architecture
Report first recorded Nov 2, 12:55
DeepSeek has released a prototype for Super-agent, an architecture enabling AI agents to maintain task context across multiple sessions and platforms. The system features a dual-layer design with a Router for task dispatch and Workers for execution, incorporating mechanisms like speculative execution and event subscription to allow continuous operation, parallel task handling, and autonomous bug detection with PR submission.
Read original source ↗
Event discussion
Reader opinions are hosted on GitHub Discussions; posting requires a GitHub account. Comments are not verified facts.