loudMore/dsh-drop-to-path
101Last commit Aug 14, 2026
dsh-drop-to-path DSH plugin
This plugin intercepts drag-and-drop and paste events in DSH's Web profile. Images keep the native attachment UI (preview/remove); other files appear as square chips. On send, all files are uploaded to the workspace and their paths are appended to the message, bypassing the image preflight check for text-only models.
How to install the dsh-drop-to-path DSH plugin
dsh plugin --profile web add /path/to/dsh-drop-to-pathCopying does not run this command. Review the repository and version before installing the dsh-drop-to-path DSH plugin.
dsh-drop-to-path DSH plugin data source
dsh-drop-to-path DSH plugin snapshot date: Aug 16, 2026
discovered
What the dsh-drop-to-path DSH plugin can do
- Intercept drag-and-drop and paste for images, documents, archives, video, and audio
- Display images with native DSH attachment cards (preview, remove)
- Display non-image files as removable square chips with format icons in the attachment rail
- Automatically upload files to workspace/.drops/ and convert them to paths on send
- Bypass DSH's image admission preflight so text-only models can receive image paths
- Support mixed batches (images + other files) uploaded in order and sent as one text block
Where the dsh-drop-to-path DSH plugin fits
- Send a screenshot to a text-only model (e.g., deepseek) by pasting it directly
- Drag a PDF into the composer and let the model process it with document tools
- Attach multiple images for pixel-diff analysis via dsh-vision-toolkit
- Drop a video file for summarization by a text-only model that uses workspace paths
- Quickly share a spreadsheet or Office document by pasting it into the chat
Who the dsh-drop-to-path DSH plugin is for
- DSH users who run text-only models (e.g., deepseek) and need to send images or files
- Users who pair this plugin with dsh-vision-toolkit to give vision capabilities to text-only models
dsh-drop-to-path DSH plugin limitations
- Relies on DSH's undocumented internal interfaces and may break after DSH upgrades (see ADAPTING.md)
- Only works with the DSH Web profile; requires Node.js and dsh CLI
- File size limits: images ≤30MB, other files ≤100MB
- Primarily designed for text-only models; multimodal models can already send images natively
dsh-drop-to-path DSH plugin: from the repository README
Quoted from the loudMore/dsh-drop-to-path README, the upstream source of the dsh-drop-to-path DSH plugin. Copyright remains with the original authors.
> Drop or paste images, PDFs, office docs, zips, videos and audio into DeepSeek Harness — they reach your text-only model as **workspace file paths**. Images keep the native attachment UI (preview / remove), everything else shows as square chips in the attachment rail, and paths are appended automatically when you hit send. [English](README.md) | [中文](README.zh.md) [](https://github.com/topics/dsh-plugin) ## Demo  *Drag images / PPT / TXT / ZIP into the composer → square file chips appear next to image thumbnails → hit send → the model receives workspace file paths.*  *The effect: images and files arrive at the text-only model as workspace paths, ready for vision / document tools.* ## Why this plugin I use a text-only model (deepseek) in DeepSeek Harness and frequently need to send screenshots to the agent. But DSH image attachments go through the model's native attachment channel, and text-only models are rejected by a preflight check before the message is even sent — the composer shows a system toast: > **The cu
Read the full READMERepository license: MIT
dsh-drop-to-path DSH plugin questions
How do I install dsh-drop-to-path?
Run `dsh plugin --profile web add /path/to/dsh-drop-to-path` in your terminal. Make sure you have Node.js and the dsh CLI installed. After installation, restart your DSH Web profile (relaunch the launcher or restart `dsh web`). No additional configuration is needed.
Does this plugin work with multimodal models?
Yes, it works but is unnecessary for multimodal models because they can already send image attachments directly. The plugin is designed to bypass the image preflight check that blocks text-only models. If you use a multimodal model, the plugin will still convert files to paths, but you may not need that feature.
Why are my files not being sent as paths?
Make sure the plugin is properly installed and the Web profile is restarted. Check that you are using a text-only model (e.g., deepseek) – if the model supports images natively, the plugin may not intercept. Also verify that the file type and size are within the supported limits (images ≤30MB, other files ≤100MB). If upload fails, a visible notice appears and the message falls back to native send without paths.
Can I paste or drag files from the clipboard?
Yes, the plugin supports both drag-and-drop and paste from the clipboard for all supported file types (images, PDFs, Office docs, archives, video, audio). When you paste or drop, non-image files appear as square chips in the attachment rail, and images show as native attachment cards.
How do I clean up the uploaded files?
Files are stored under `<workspace>/.drops/` in your DSH workspace directory. You can delete them manually anytime. The plugin does not automatically clean up old files, so you may want to periodically remove them to free up space.