Introduction¶
DeepSeek Harness (DSH) currently mainly handles text or API calls. In the image processing domain, local tools such as Adobe Photoshop remain powerful, but lack automated interfaces. This plugin solves the problem of “AI being unable to directly operate local GUI applications.”
What It Is¶
This is a DSH plugin maintained by Better-Rain. Its core value is enabling the AI in DSH to directly drive your local Adobe Photoshop. You can control Photoshop with natural language to perform batch background removal, batch processing, or run arbitrary scripts, without manually handling each image one by one.
Core Features¶
The plugin provides the following capabilities:
1. Environment self-check: Checks the Photoshop version and command availability.
2. Read current screen state: Retrieves the complete layer tree, supporting both index-based paths and name-based paths.
3. Batch background removal: Supports subject selection and background removal, outputting PNGs with transparency.
4. Batch production: Applies the same set of operations to a batch of files.
5. General operations: Includes 93 operations, covering documents, layers, adjustments, filters, selections, and more.
Installation and Activation¶
Run the following command in the terminal to install:
dsh plugin --profile web add github:better-rain/dsh-plugin-photoshop
After installation, restart dsh web for the changes to take effect.
Typical Usage¶
- Extract the person from all images in
D:\framesand save the results toD:\cutouts. - Use
photoshop_applywithdry_run: trueto preview what it will affect.
Applicable Scenarios and Notes¶
The plugin runs with the permissions of the current dsh process. Please review the source code and license before installing.
- Runtime environment: It must run on Windows, and Photoshop must be installed locally.
- Privacy and security: The plugin does not connect to the internet, does not upload any data, and all processing is completed locally.
- Dependencies: Zero third-party dependencies.
Conclusion¶
This plugin is suitable for scenarios that require automated image processing in DSH or batch execution of Photoshop operations. For more details, see GitHub.