Show HN: I made a skill to tell AI on how to use human as Dobby

Summary

This is a prompt skill for assigning simple support tasks to a human helper called "Dobby." It frames the human as a low-skill executor for chores such as file moving, manual testing, information gathering, formatting, and copy-paste work.

Problem

The project targets a coordination problem: how an AI or operator should give simple, bounded tasks to a human helper.
It matters for human-AI workflows because many software tasks still need manual execution, browsing, testing, or formatting outside the model's direct control.
The excerpt does not define a measured research gap, formal task, or evaluation setting.

Approach

The core method is a reusable "Skill" that users install with npx skills add littlelittlecloud/dobby.
The skill gives behavioral instructions for managing a human assistant named Dobby and sets a sharp task boundary: Dobby handles simple physical or clerical work, while coding and architecture stay with the main agent or user.
The examples of allowed work are concrete: moving files, manual testing, collecting materials, organizing formats, and copying and pasting.
The mechanism is prompt-level role assignment rather than a trained model, new algorithm, or system with autonomous planning.

Results

No quantitative results are provided in the excerpt.
No benchmark, dataset, accuracy, task success rate, latency, or comparison against a baseline is reported.
The strongest concrete claim is functional scope: the skill is meant to help an AI instruct a human to do simple execution tasks and avoid assigning high-intelligence work such as coding or architecture design.
The artifact appears to be installable through npx skills add littlelittlecloud/dobby, but the excerpt gives no usage metrics or outcome data.

Link

https://github.com/LittleLittleCloud/Dobby