AI Agent Hub
Back to plugins
dsh-agent-message preview

dsh-agent-message

Workflow Updated 2026.08.19

Run the following command in DeepSeek Harness:

dsh plugin install GengDaPeng/dsh-agent-message

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

To install this plugin in DeepSeek Harness, run the command 'dsh plugin install GengDaPeng/dsh-agent-message', with source code available at https://github.com/GengDaPeng/dsh-agent-message.

About this plugin

In the multi-Agent environment of DeepSeek Harness, multiple sessions often run concurrently within a single process, but they typically lack direct communication channels, making orchestration, collaboration, or supervision tasks cumbersome. The dsh-agent-message plugin addresses this by enabling cross-session communication for each Agent session, allowing different sessions to interact like sending messages. Its core capabilities include listing all sendable independent sessions, delivering messages to target sessions (with modes like followup, steer, and inject), and checking delivery status on demand, such as pending or claimed. The plugin also integrates intuitive @ session targeting and visible message cards, simplifying user operations for specifying recipients and viewing communications. This plugin is ideal for developers who need to orchestrate, collaborate, or supervise multiple Agents—for example, an orchestrator Agent assigning tasks to a developer Agent, two Agents working in relay, or a supervisor Agent monitoring worker session activity. By providing standardized tools, it streamlines multi-Agent workflow management, making communication reliable and transparent.

Screenshots

Use Cases

  • An orchestrator Agent assigns coding tasks to a developer Agent
  • Two Agents collaborate in relay to complete a complex project
  • A supervisor Agent monitors the work status of multiple workers in real-time

Best For

  • AI developers building multi-Agent systems
  • Engineers automating workflow orchestration
  • Technologists managing and supervising Agent tasks