AI Agent News/stories/deepseek-open-sources-dsh-superagent-a-router-worker-coding-agent.md

Frameworks

DeepSeek open-sources dsh-superagent, a router-worker coding agent

DeepSeek has released dsh-superagent, an MIT-licensed research prototype for coding agents. It turns requests from GitHub and Feishu into parallel worker sessions.

Source: www.marktechpost.com/2026/11/01/deepseek-dsh-superagent-router-worker-coding-agent

DeepSeek released dsh-superagent under the MIT license. The project is described as a research prototype rather than a production product. It takes requests arriving from GitHub and Feishu and turns each one into parallel worker sessions, with a router-worker structure.

In DeepSeek's own simulation, p95 task completion fell to 15.7 minutes from 29.5 minutes. The figures come from that simulation, so they are not measurements from live deployments. The MarkTechPost write-up does not describe benchmark conditions in further detail.

Why it matters if you build agents

Developers building coding agents can study a router that fans incoming GitHub and Feishu requests out to parallel worker sessions, since the code is MIT-licensed. The reported latency drop is a simulation result, so it should be checked against real workloads before it shapes design choices.

Get stories like this every weekday

One short email with the day's agent tooling news.

Subscribe