Introduction

The core capability of DSH lies in extension through plugins. When you need to execute a set of tool calls repeatedly and unattended, calling tools directly in a local conversation is inefficient and unreliable. dsh-lines aims to solve this scenario: it solidifies a locally verified, successful tool-call sequence into a named pipeline, which is hosted on a server and runs on a schedule. Each run generates a verifiable “receipt.” It also supports creating monitor tasks that notify you only when results change.

Core Capabilities

This plugin provides the following core functions for managing pipelines and tasks:

  • what_can_you_do: Describe a task in natural language (any language) and receive the corresponding server tools and how to call them.
  • create_pipeline: Convert the current tool-call sequence into a reusable pipeline.
  • run_pipeline: Execute the pipeline immediately and retrieve the result and credentials for each step.
  • create_task: Create a scheduled task in the cloud without keeping a local process running, and get notified only when results change.
  • list_pipelines / list_tasks: View the list of pipelines and tasks.
  • get_task_runs: Retrieve the run history of a task.
  • delete_pipeline / delete_task: Delete a pipeline or task.

Installation

Run the following command in your terminal to install it:

dsh plugin --profile <your-profile> add github:mario03690/dsh-lines

Usage Flow

  1. Describe the task: Use what_can_you_do to describe the task you want to complete and obtain the corresponding server tools and calling methods.
  2. Solidify the pipeline: Use create_pipeline to convert the tool-call sequence into a named pipeline.
  3. Execute or schedule it: Use run_pipeline to run it immediately, or use create_task to set up scheduled execution.

Security and Limitations

  • Permission scope: The plugin runs in the DSH v0.1 developer preview environment. It includes no local tool code and does not involve filesystem or Shell/process access. It only uses an outbound HTTPS connection to ainetcafe.com.
  • Data privacy: Documents and prompts are processed in memory and are not retained. The ?s=dsh-lines parameter in the configuration URL is only an identifier for the installation path and is not used for tracking.
  • Dependencies: The only dependency is @deepseek-ai/dsh-mcp-client (installed with DSH).

Cost and Quotas

  • Free quota: You can use the free anonymous quota without registration. The quota covers the first 7 days or 60 calls, whichever comes first.
  • Not billed: Failed run self-checks are not counted toward your usage.
  • Monitor tasks: Monitor tasks (such as a daily China reachability check) are free.
  • Paid: Subscribe after exceeding the quota. An optional AllRouter key is sent only as a request header and is not stored.

Conclusion

dsh-lines is suitable for scenarios where you need to stabilize and automate workflows. It turns loose sequences of tool calls into rigorous cloud pipelines and provides auditable run records. The full documentation and source code are available at GitHub.