dsh-plugin-zai-mcp
Run the following command in DeepSeek Harness:
dsh plugin install nekomona/dsh-plugin-zai-mcp
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install nekomona/dsh-plugin-zai-mcp in the DeepSeek Harness terminal to install; source repo at https://github.com/nekomona/dsh-plugin-zai-mcp. Restart DSH to activate.
About this plugin
DeepSeek Harness users who already have a z.ai or Zhipu GLM model wired in often miss a perception layer: reading screenshots, understanding technical diagrams, pulling the latest web results, or browsing open-source repo structures. dsh-plugin-zai-mcp bridges all four z.ai GLM Coding Plan MCP services (vision, web reader, web search, zread) into native DSH tools in a single install, exposing 13 tools that span everything from analyze_image and understand_technical_diagram to webSearch and read_file.
If your GLM Coding Plan key already lives in DSH's managed credentials, the plugin works out of the box with zero extra configuration: it resolves the key automatically, and no environment variables or manual endpoint wiring are needed. The architecture is fail-soft by design - an unreachable endpoint or a missing key degrades to per-call tool errors for that service only, leaving the other three bridges and the DSH host process untouched.
This plugin is aimed at developers who work with z.ai or Zhipu GLM models inside their daily DSH workflow and want their AI assistant to go beyond plain text - understanding images, videos, web pages, and repository layouts - without juggling multiple MCP endpoints, hand-managed keys, and per-service timeout settings.
Use Cases
- Let the AI assistant read screenshots, technical diagrams, and data visualizations on the fly
- Perform real-time web search and full-page reading inside a conversation
- Explore open-source repository structures and pinpoint specific files
Best For
- Developers who already run z.ai or Zhipu GLM models inside DSH
- Engineering teams that need their AI assistant to understand images and videos
- Users who want to bridge multiple MCP endpoints without duplicating key configuration
Related Plugins
A unified suite combining hot runtime injection, task-aware thinking-mode routing, and a graded session protocol with red-team gates to sustain model diligence across long-horizon inference.
ModLens is a vision plugin for DeepSeek Harness that gives text-only models sight by reading images pasted directly into chat, with zero-config setup and multiple vision engines.
On-demand vision for text-only DeepSeek Harness agents: built-in free keyless vision chain and 14 vision tools, routing image turns as tool calls to vision models with pixel fidelity, no Python needed, one-command install.
Give text-only models in DeepSeek Harness eyes, enabling image Q&A, long-screenshot OCR, UI restoration, and GUI visual tasks.