AI Agent Hub
Back to plugins
🧰

dsh-rich-artifacts

Web Tools Updated 2026.08.17

Run the following command in DeepSeek Harness:

dsh plugin install Inkotake/dsh-rich-artifacts

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

Install the plugin in DeepSeek Harness by running dsh plugin install Inkotake/dsh-rich-artifacts; the source is hosted at https://github.com/Inkotake/dsh-rich-artifacts .

About this plugin

Anyone who has asked a model to produce an image or a report knows the awkward moment that follows: the model says "saved to output/report.png" and hands you a file:/// link. You stitch the path together by hand, open a browser, and the experience breaks at an invisible wall. dsh-rich-artifacts removes that wall. When the model creates a user-facing file, it calls the publish_artifact tool and the file appears right in the conversation tail — raster images rendered inline, everything else as a clean download card.

Under the hood, artifacts are stored as SHA-256 content-addressed blobs with metadata kept separately, so historical sessions remain fully browsable after reload or fork. Security is handled in layers: workspace path containment, symlink rejection, and magic-byte MIME sniffing that prevents a file disguised as a .png from actually inlining as HTML. SVG and HTML are strictly download-only in V0.1, and hard caps are enforced on file size, image dimensions, and pixel count. PNG, JPEG, WebP, and GIF render inline; all other formats become file cards with a download button.

It is built for DeepSeek Harness developers who frequently produce visual deliverables — data charts, design mockups, PDF reports, CSV datasets — and want to turn "the file is on disk" into "the file is right here." Installing this plugin removes the manual step and adds a durable, session-persistent delivery path for every artifact the model produces.

Use Cases

  • Preview chart images inline in the conversation right after the model generates them
  • Deliver CSV or PDF files as download cards without manual path handling
  • Keep published artifacts visible and downloadable after session reload or fork

Best For

  • Developers building AI workflows on DeepSeek Harness
  • Users who need the model to frequently produce charts, reports, or other visual deliverables
  • Team automating scenarios where manual file handling is a friction point