The philosophy behind DeepSeek Harness (DSH) is “everything is a plugin.” When using large language models to perform specific tasks, the output is often disorganized or contains excessive redundant information. To address this pain point, you can use the ADHD-friendly output plugin to constrain model behavior so that its output aligns better with reading habits.
Plugin Positioning¶
This plugin is maintained by user satan9394. It is a DSH client plugin designed to solve problems such as disorganized output structure, off-topic content, and excessive pleasantries through mandatory output format constraints.
Core Features¶
The plugin mainly includes the following capabilities:
- Action first: Prioritize concrete action steps.
- Numbered steps: Number the operation workflow.
- Off-topic suppression: Prevent the model from drifting away from the core question.
- State restatement: Clearly restate the current status in the response.
- Specific time estimates: Add a time dimension to task descriptions.
- Lists ≤ 5 items: Limit the length of output lists and avoid information overload.
- No pleasantries: Remove unnecessary greetings and polite filler.
Installation and Activation¶
The official installation command cannot currently be confirmed. Please visit the GitHub repository to view the source code and specific configuration details.
Typical Usage¶
Because detailed usage examples cannot be confirmed, usage should rely on the core feature characteristics mentioned above:
- When making a plan, the model will prioritize providing an action checklist.
- For multi-step tasks, the output will include a numbered list.
- The number of list items is limited to no more than 5.
Applicable Scenarios and Notes¶
This plugin is suitable for scenarios that require quickly obtaining execution steps or status reports. The plugin runs with the permissions of the current DSH process. Before installation, review the source code and license (currently unconfirmed).
Summary¶
This plugin solves messy output through format constraints. For detailed information, see the DSH directory or the GitHub repository.