Ultrathink Plugin
A simple OpenCode plugin that automatically injects "Ultrathink: " before user prompts for GLM-4.6 and Big Pickle models, encouraging the AI to engage in deeper, more thoughtful reasoning.
What It Does
The Ultrathink plugin modifies your outgoing messages to AI services by prepending "Ultrathink: " to each user prompt only when using GLM-4.6 or Big Pickle models. This simple prefix can help elicit more detailed, analytical responses from these specific AI models by signaling that you want them to engage in deeper thinking about your request.
Installation
Add to your opencode.json or .opencode/config:
{
"plugin": ["opencode-ultrathink-plugin"]
}
How It Works
- Model-specific: Only activates for GLM-4.6 (
glm-4.6) and Big Pickle (big-pickle) models
- Intercepts API calls: Monitors fetch requests to AI providers
- Injects prefix: Automatically adds "Ultrathink: " to the beginning of your user messages for supported models
- Format-agnostic: Works with different API formats while preserving message structure
- Transparent: No changes to your workflow - just better AI responses when using supported models
Example
Without the plugin:
User: Explain quantum computing
With the plugin:
User: Ultrathink: Explain quantum computing
Acknowledgments
This plugin was created using the Dynamic Context Pruning plugin by Dan Tarquinen as a template. The DCP plugin provided the excellent fetch wrapper architecture and API format handling that made this implementation possible.