AI Agent Hub
Back to plugins
🖥️

dsh-postman

Client Updated 2026.08.25

Run the following command in DeepSeek Harness:

dsh plugin install zhousun55-byte/dsh-postman

Paste the following prompt into your AI chat to install this plugin:

Run dsh plugin install zhousun55-byte/dsh-postman in the DeepSeek Harness terminal; the source is available at https://github.com/zhousun55-byte/dsh-postman . Refresh the page after installation to start using the plugin.

About this plugin

Referencing local files in a DSH conversation used to mean copying content into a workspace and typing out paths, which is clunky and error-prone. dsh-postman (Postman) puts file upload right inside the input box type /upload in the command menu and pick files or a whole folder with no extra buttons or intermediate steps.

Images go through the native DSH attachment pipeline, appearing as thumbnails and image blocks in the message so the vision bridge transcribes them just like a pasted image. Text files are injected as inline code blocks in the draft for immediate model access, while other binaries get their path and size for the read tool. Folder uploads preserve the directory tree on disk, auto-create directories, and refuse to overwrite on name conflicts. The draft receives a file list plus text previews, and images inside the folder (up to 12) are added as image blocks too. The plugin is zero-dependency and build-free, using only Node built-ins.

Best for developers who frequently reference local images or documents in conversation, or anyone who wants to batch-upload a directory structure for the model to process in one go. No workspace changes, no path memorization just drop files into the conversation and go.

Use Cases

  • Upload local images directly into the conversation for model analysis
  • Batch-upload a project folder so the model understands the full codebase structure
  • Reference local documents in chat without copying file paths

Best For

  • Developers who frequently reference local files in DSH conversations
  • Teams that need to batch-upload directory structures for holistic model understanding
  • Users wanting a zero-config workflow without manual workspace operations