Who it is forA developer who keeps a dsh machine running 24/7 and wants to file work from elsewhere without watching each run.
The problemEach task normally waits for a confirmation, which blocks unattended operation on a server.
The outcomeTasks you file run on schedule in the sandbox at off-peak times and the report is written with no confirmations.
What you need first
- DeepSeek Harness (dsh) installed
- The lowtide plugin installed into dsh
- A model connected to your Harness
- A workspace directory for the project the tasks will touch
Steps
- In settings, switch the autonomy level to L3 full-auto, accepting the double confirmation.
- Confirm the workspace and the sandbox preset you want the tasks to use.
- Set the daily budget in settings so unattended spending stays within your limit.
- File tasks through the API route
POST /ds-lowtide/tasks. - Check the execution report later to confirm the tasks ran and the budget was respected.
Prompt for your agent
Using the lowtide plugin in DeepSeek Harness on a machine that runs 24/7, switch autonomy to L3 full-auto and file the task <task description> through POST /ds-lowtide/tasks. Expected output: the task runs at off-peak and appears in the execution report. Verify that the task status reaches done or a clear failure reason, and that the daily budget was not exceeded.
Caveat. The README says the HTTP routes accept only same-origin requests from the same machine, which may block filing tasks from another location despite the remote-filing example.
Use this in your AI agent
Ask your agent
Using the Deduped MCP, get the setup steps for KelaoHu/dsh-lowtide and "Run queued tasks unattended on an always-on dsh server"
Add Deduped to Claude Code once
claude mcp add --transport http deduped https://dedupedai.com/mcp
Every Monday: the 5 hottest repos and exactly how to use them. Free, one email a week, unsubscribe anytime.
Get them freeShare: X LinkedIn Hacker News Email Copy link
Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer