AI Agent Hub
Back to plugins
🖥️

dsh-intercom

Client Updated 2026.09.11

Run the following command in DeepSeek Harness:

dsh plugin install Sqhao-O/dsh-intercom

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

Paste this prompt into a running DeepSeek Harness session to install the plugin (source: https://github.com/Sqhao-O/dsh-intercom): Install the plugin from github:Sqhao-O/dsh-intercom, restart dsh, and verify that the intercom tool is registered and available in every session.

About this plugin

Running multiple dsh sessions side by side on the same machine—say a planner, a worker, and a reviewer—yet having no built-in way for them to talk to each other is the most immediate friction in local multi-agent workflows. dsh-intercom solves exactly that: it auto-spawns a lightweight local broker (via a unix socket or a Windows named pipe) so sessions in any two independent dsh processes can discover each other and exchange messages, with no cross-host networking or external message bus required.

Capabilities span three layers. Discovery: list and list-cwd show every live session on the machine with its name, working directory, model, and real-time state (idle, thinking, tool). Messaging: send is fire-and-forget, with mailbox queueing so a message to a briefly offline named peer is delivered automatically when that peer reconnects; ask/reply is a blocking request-reply that guarantees the sender waits for an answer before continuing. Addressing: each session can take a stable alias, and peers resolve by alias, full session id, or unique prefix. Installing the plugin also registers a coordination skill automatically, offering planner-worker playbooks and send-versus-ask guidance with zero extra steps. In the web profile, a dedicated Intercom page in the Settings panel lists the live roster and lets you send a message as any session hosted by that process.

This plugin is for developers who orchestrate multi-agent workflows on a single workstation: when a planner needs to dispatch a task to a worker, the worker needs to ask the planner for the next step, or sessions need to pass context back and forth, this local interconnect makes multi-session collaboration feel as natural as in-process thread communication—without pulling in Redis, RabbitMQ, or a sidecar HTTP service.

Use Cases

  • A planner session dispatches a task to a worker session and blocks until a reply arrives
  • Multiple dsh sessions on the same machine discover each other and exchange context
  • Use the Web panel to browse the local session roster and send a message as a chosen session

Best For

  • Developers orchestrating multi-agent workflows on a single workstation
  • Teams needing local inter-session messaging without an external message bus
  • Engineers debugging multi-session pipelines via the dsh web profile