Updated

dsh-superagent is a research prototype from the DeepSeek Harness team of a proactive coding agent that serves one GitHub repository.

Pricing Free open source
Visit site

Follow dsh-superagent to get an email when its listing changes, such as new pricing or features. One digest a day at most, and you can unsubscribe anytime.

Company DeepSeek
Pricing model Open Source
Paid plans None listed
Category peers 53
01Overview

It listens to GitHub and optionally Feishu, decides under a team-written Policy whether a request is authorized, runs each authorized request as a Ticket in its own DSH session, and replies in the original conversation, for example with a pull request.

02Key features
  1. 01

    Packaged as a DeepSeek Harness bundle that listens to GitHub issues and pull requests, with optional Feishu chat

  2. 02

    Eval gate with allowed-sender lists, admission rules and per-Ticket Mandates that messages cannot override

  3. 03

    Each authorized request becomes a Ticket executed in its own DSH session and working directory

  4. 04

    Superagent web panel with Overview, Source, Trigger, Eval, Execution and Settings tabs that surfaces work needing a person

  5. 05

    Agent-guided setup: a coding agent follows SETUP.md and asks the operator one question at a time

03Pricing

Open Source

dsh-superagent is free and open-source under the MIT license. It runs on your own machine inside DeepSeek Harness, so the only costs are the model provider you configure for DSH and any GitHub or Feishu accounts you connect.

04Who it's for

Best for

Teams already experimenting with DeepSeek Harness who want an always-on agent that picks up authorized requests from GitHub and answers with pull requests, and who want to study DeepSeek's super-agent design.

Not ideal for

Production use or teams wanting a supported product: DeepSeek labels it research code that may contain bugs, it covers a single repository, and it depends on alpha DSH packages plus Node.js, pnpm, tmux and the GitHub CLI.

05Verdict
A rare look at how a frontier lab runs an autonomous, attention-saving coding agent internally, worth studying and trying with DeepSeek Harness, but explicitly a prototype rather than something to depend on.
ai.dosa.dev editorial · Nov 1, 2026
06Recent updates

Published on 2026-10-31 as version 0.1.0 alongside DeepSeek's article "Exploring Super-agent Design: Towards a More Autonomous Agent System"; it gathered more than 6,700 GitHub stars within its first day.

As of

07Pricing & FAQ
Is dsh-superagent free?

Yes - dsh-superagent is Open Source. dsh-superagent is free and open-source under the MIT license. It runs on your own machine inside DeepSeek Harness, so the only costs are the model provider you configure for DSH and any GitHub or Feishu accounts you connect.

Is dsh-superagent open source?

Yes - dsh-superagent is open source. dsh-superagent is free and open-source under the MIT license. It runs on your own machine inside DeepSeek Harness, so the only costs are the model provider you configure for DSH and any GitHub or Feishu accounts you connect.

Who is dsh-superagent best for?

Teams already experimenting with DeepSeek Harness who want an always-on agent that picks up authorized requests from GitHub and answers with pull requests, and who want to study DeepSeek's super-agent design.

Who is dsh-superagent not ideal for?

Production use or teams wanting a supported product: DeepSeek labels it research code that may contain bugs, it covers a single repository, and it depends on alpha DSH packages plus Node.js, pnpm, tmux and the GitHub CLI.

What are the best dsh-superagent alternatives?

The closest dsh-superagent alternatives on ai.dosa.dev are Jules by Google, OpenHands, SWE-agent - all listed under Autonomous & Async Agents.

Who makes dsh-superagent?

dsh-superagent is developed by DeepSeek. It is listed in the Autonomous & Async Agents category on ai.dosa.dev.

08More in Async Agents
09Tags
  • Autonomous & Async Agents
  • DeepSeek Harness
  • GitHub Automation
  • AI Agents
  • Open Source
Track dsh-superagent in your AI stack

Favorite this tool to revisit it later, or Zap it to contribute to the public vote count.

Compare Async Agents Sign in to save
Content on this page is AI-generated. Please verify details with the vendor's website for accuracy. Are you the maker? Get your Featured badge ← Back to directory