dsh-agent-team-model
Run the following command in DeepSeek Harness:
dsh plugin install XMeowchan/dsh-agent-team-model
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install XMeowchan/dsh-agent-team-model in DeepSeek Harness to install the plugin; source code is available at https://github.com/XMeowchan/dsh-agent-team-model .
About this plugin
In DeepSeek Harness Agent Teams, every teammate inherits the captain model and routing by default, making it hard to assign different providers, models, or reasoning efforts per role. This plugin turns that routing into a visual setting: pick a default provider, model, and reasoning strength, or override specific teammates by name, and newly spawned agents pick up the rules automatically. No code changes, no manual parameter injection.
A live team-status panel in the session shows the captain, each member, their run state, and the exact route actually in effect, so you can verify that every override landed correctly. Reasoning effort can be set to a model's own default or to inherit the captain; choosing the captain's provider means no route is injected at all. Config is persisted to a local JSON file and restored on restart without waking any agents.
Built for developers already running DeepSeek Harness with the official Agent Teams layer who want to mix providers across a team, give planners heavier models while keeping executors lightweight, or simply stop hard-coding routing into prompts.
Screenshots
Use Cases
- Assign high-reasoning models to planners and lightweight models to executors by role
- Mix multiple providers so different teammates hit different LLM backends
- Inspect the effective routing and run state of every teammate in real time during debugging
Best For
- Developers running DeepSeek Harness with the official Agent Teams layer
- Lead engineers who need per-role model routing in multi-agent teams
- Users who prefer a visual settings panel over manually injecting routing parameters
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.