AI Agent Hub
Back to plugins
🤖

dsh-bigmodel-catalog

Model Inference Updated 2026.08.25

Run the following command in DeepSeek Harness:

dsh plugin install OpenSaozi/dsh-bigmodel-catalog

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

Run dsh plugin install OpenSaozi/dsh-bigmodel-catalog in your DeepSeek Harness terminal to install this plugin; the full source is at https://github.com/OpenSaozi/dsh-bigmodel-catalog .

About this plugin

If you hold a Zhipu AI GLM Coding Plan membership but find that DeepSeek Harness does not natively expose the 1M-token glm-5.3 model in its selector, this plugin fills that gap. It registers a dedicated zai adapter that wires the Coding Plan PaaS endpoint directly into the Harness provider pipeline, making glm-5.3 (1,000,000-token context, 128K max output) alongside glm-4.5, glm-4.6, and glm-5 immediately selectable—without manual request stitching or external provider configuration.

The plugin is designed with security and self-containment in mind. The BIGMODEL_API_KEY is resolved per-request through the Harness credential service and never leaks into Cordis configuration or logs. Live endpoint discovery is paired with a verified-exception table so that glm-5.3 remains available even when the Coding Plan API does not list it explicitly. Catalog, filtering, credential resolution, and adapter registration all live inside the package; deployments can override baseURL or apiKeyEnv to target a different Coding Plan endpoint without modifying plugin internals.

It is built for developers who carry a Zhipu GLM Coding Plan or BigModel account and need to analyze large codebases, process lengthy documents, or perform intensive code generation within the Harness workflow—all while benefiting from a near-native 1M-context experience without swapping out their entire toolchain.

Use Cases

  • Select glm-5.3 in Harness to analyze codebases exceeding one million tokens
  • Leverage the 128K max output to process lengthy documents inside the Harness workflow
  • Run intensive code generation tasks within Harness via the dedicated zai adapter

Best For

  • Developers with a Zhipu AI GLM Coding Plan membership
  • Engineers holding a BigModel account who need long-context inference
  • Teams that want to call GLM models directly within their existing Harness setup