dsh-image

Has faultsCurrentUIv0.2.0Vision

Registers deepseek-vision, a provider that declares image support and appears as "multimodal LLMs", turning pictures into text before handing them to text-only DeepSeek. The multimodal provider and model used for that conversion are configurable in settings.

Install

dsh plugin --profile web add dsh-image

Swap web for your own profile name. First use initializes the profile for you.

Checks

declares-bundle
No dsh.bundle.patch — contributes no profile layer
entry-shipped
Entry module ships: lib/index.js
prebuilt
Prebuilt — installs without running build scripts
client-half-shipped
Browser half ships: lib/client.js

Version compatibility

@deepseek-ai/schemastery^3.18.1→ 3.18.1Current

@deepseek-ai/dsh-settings^0.1.0-rc.6→ 0.1.0-rc.6Current

dsh moved 0.0.1-rc.1 to 0.1.0-rc.6 in three days. Note the dsh library packages promote current releases on the next tag; latest still points at the first-day version.