X postLeandro von WerraAnnouncement
As coding agents proliferate, bottlenecks shift from writing code to reviewing and reconciling pull requests. The prototype provides useful patterns for running autonomous agents without overwhelming developer attention.
With dsh-superagent, DeepSeek has open-sourced an experimental framework exploring autonomous software engineering workflows. Designed to operate across a single GitHub repository, it listens for triggers from authorized users via issues, PRs, or Feishu conversations.
The framework prioritises task isolation and operational quietness. Each incoming task is assigned an independent agent session and directory. Rather than constantly prompting operators, the agent only acts when authorized and only reports back when results—such as completed pull requests—or blocking human decisions are reached. A web interface provides an overview of work items that specifically require human review.
The maintainers explicitly describe the repository as research code originating from internal day-to-day work, largely engineered with the assistance of coding agents. It lacks the polish of a commercial release and does not incorporate all architectural ideas outlined in DeepSeek's theoretical publications. Teams examining the tool must treat it as an experimental foundation that requires rigorous security and review policies before internal deployment.
Leandro von Werra
@lvwerra · 31 October 2026
Another DeepSeek drop: dsh-superagent An agent framework that works through a repo's GitHub issues and PRs with many agents in parallel, designed to take as little human attention as possible while delivering as much value as possible. t.co/3H2v9z6Wn6 t.co/E1B75oOkEJ
2.4k likes122 replies2.5k bookmarks166.9k views
View on XSummary generated by AI (Gemini 3.8 Flash, 2 November 2026). It may contain errors – the original source is authoritative.
Similar in content, found by the AI search.
Repository:deepseek-ai/dsh-superagent-prototype
DeepSeek has published dsh-superagent, an experimental proactive coding agent tailored to individual GitHub repositories. Packaged for DeepSeek Harness, it handles development tasks in dedicated workspaces and integrates optional Feishu support. The system is designed to provide actionable development results while demanding minimal human intervention.
9kstarsTypeScript
Link:LangChain
LangChain introduces Deep Agents, an open source agent harness built for long-running, autonomous workflows with native planning, context handling, and subagent orchestration.
Repository:deepseek-ai/deepseek-harness
DeepSeek Harness (dsh) is an open-source AI agent harness by DeepSeek AI. Built with TypeScript and Cordis, it uses an 'everything-is-a-plugin' architecture to provide modular agent capabilities and a Web UI.
212.8kstarsTypeScript
Link:superset.sh
Superset is a developer workspace designed to orchestrate multiple CLI coding agents, such as Claude Code, Codex, and OpenCode, in parallel. By isolating tasks inside separate Git worktrees, it prevents file collisions while allowing developers to monitor, review, and merge code changes from a single interface.
Link:DeepSeek
DeepSeek has launched a developer preview of DeepSeek Harness, an open-source runtime framework for AI agents. Built on the Cordis kernel, it implements an architecture where all capabilities are plugins alongside an append-only event stream for full traceability.
Repository:PrimeIntellect-ai/prime-agent
Prime Agent is an open-source coding and research agent designed for complex and long-running autonomous workflows. It combines a Recursive Language Model (RLM) architecture with a Continual Harness to manage context programmatically and persist learned skills across sessions.
19.9kstarsTypeScript
Parts of this content were created with the assistance of AI.
