DeepSeek releases Super-Agent, an open coding agent that reads team chats
DeepSeek put Super-Agent on GitHub under an MIT licence. It takes coding tasks from team chat and answers with pull requests; the lab calls it a research prototype.
Source published · In the Daily Brief of 2 November 2026
Filed under Agents, AI industry
DeepSeek published the code for Super-Agent, an experimental agent that takes on coding work for a software team, under the name dsh-superagent and an MIT licence. The team behind it set out the design in a WeChat post on Saturday. Bloomberg reported that the project is aimed at Anthropic’s Claude Code.
The agent works on one GitHub repository and can listen to Feishu, ByteDance’s workplace chat app. When an approved person asks for work, it opens a task in its own session and replies in the original thread, for example with a pull request. A web panel lists what needs a human. The stated aim is to take as little human attention as possible.
DeepSeek calls it a research prototype and says the public code differs from its internal version, with some ideas not yet built. In January the CooperBench study found coding agents’ success rates were on average 30% lower working as a team than alone; it did not test this agent.