#162162nd of 389 in Skills, plugins and prompts
dsh-agent-teams
Plugin that lets a DeepSeek Harness session run a team of sub-agents
- Stars
- 2.9k
- License
- MIT
- Last commit
- Nov 2026
- Last release
- Nov 2026
- Language
- JavaScript
13 on 11 Aug 20262,868 on 10 Nov 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
dsh-agent-teams is a DeepSeek Harness plugin that turns the current session into a captain. The captain creates durable sub-agent members, splits a goal into dependency-aware tasks, and coordinates them through mailbox messages. It adds 14 captain tools, file-backed state under .agent-teams/, a shared-task scheduler, a plan-approval step and a Web UI with a task DAG. A /agent-teams slash command starts it.
Who it is for: DeepSeek Harness users who want multi-agent task delegation inside one session
Strengths
- Plans a staged roster and task DAG for approval before any member starts
- Idle members automatically claim ready tasks; reassignment revokes stale attempts
- Per-member provider/model routing and owner-defined tool policies via config
- Team state persisted to disk, with archives of completed teams
Weaknesses
- Requires DeepSeek Harness; the recommended host version is still a release candidate
- One captain leads one active team at a time
- State is file-backed; concurrent processes editing the same team are not coordinated
- Tool policies limit model tool visibility only, not an OS sandbox
What it needs
- Needs DeepSeek Harness (@deepseek-ai/dsh), Node.js/npm
- Models: DeepSeek Harness-configured providers and models (e.g. deepseek-v4)
Quick answers
Is dsh-agent-teams open source?
Yes. Its code is on GitHub under the MIT license.
Also in Skills, plugins and prompts
See all 389| Rank | Project | Score |
|---|---|---|
| 1 | claude-memPersistent cross-session memory plugin for Claude Code and other coding agents | 97 out of 100 |
| 2 | impeccableDesign skill with 24 slash commands for AI coding agents' frontend output | 97 out of 100 |
| 3 | gstackMarkdown slash-command skills that give Claude Code planning, review, QA and ship roles | 96 out of 100 |
| #4 | rufloAgent harness adding swarms, memory and hooks to Claude Code and Codex | 96 out of 100 |
| #5 | cavemanShortens coding-agent replies and compresses tool output before the model reads it | 95 out of 100 |