AI Agent Hub
Back to plugins
dsh-plugin-agent-teams preview

dsh-plugin-agent-teams

Client Updated 2026.08.22

Run the following command in DeepSeek Harness:

dsh plugin install Wuxie233/dsh-plugin-agent-teams

Paste the following prompt into your AI chat to install this plugin:

Run dsh plugin install Wuxie233/dsh-plugin-agent-teams inside DeepSeek Harness to install the plugin; the full source is available at https://github.com/Wuxie233/dsh-plugin-agent-teams . Restart DSH after installation to activate AgentTeams.

About this plugin

A single DeepSeek Harness chat can only go so far when a task demands parallel review, analysis, and implementation. dsh-agent-teams turns the active session into a small, commandable crew: you describe the goal in plain language, and the plugin assembles roles, splits the work into a dependency-aware task graph, coordinates multiple members moving in parallel, and hands back one consolidated deliverable through the captain.

The plugin provides ten coordination tools built around five core capabilities. The captain creates the team, assigns role-specific duties, and merges the final output. Every member is a durable, resumable DSH sub-agent that can be woken up for focused follow-up turns. Tasks flow through explicit states and cannot be claimed until their dependencies are satisfied. Members exchange persistent mailbox messages directly with teammates or the captain, with no central relay. A live Web activity panel surfaces roles, current work, unread messages, task dependencies, and archived team history at a glance. No separate workflow engine is required; state is file-backed, and member runtime snapshots inherit the captain's current model and reasoning settings automatically, so the default path needs zero extra configuration.

It is built for developers who want to drive multi-role collaboration with a single sentence: a commit review that needs performance, security, and product perspectives at once; a feature build that can be split across backend, frontend, and test streams in parallel; or any scenario where work must be divided, advanced independently, and reassembled into one coherent result. Hand the brief to AgentTeams and let the crew take it from there.

Screenshots

Use Cases

  • Kick off a commit review from performance, security, and product perspectives in parallel, then have the captain merge one consolidated report
  • Split a feature build across backend, frontend, and test members in parallel, unlocking follow-up tasks only after their dependencies finish
  • Describe a goal in one natural-language sentence and let multiple sub-agents divide the research, drafting, and QA work before the captain assembles the final deliverable

Best For

  • Individual developers already on DeepSeek Harness who want a single session to drive multiple roles in parallel
  • Small teams that need to fan out reviews, analysis, and implementation to different perspectives at the same time
  • AI-workflow users who prefer a natural-language entry point and do not want to deploy a separate workflow engine