Overview
Executes LLM instructions using the workspace agent’s active LLM model. Supports variable interpolation to customize prompts dynamically.
Input Variables
Instructions: The prompt or directives sent to the LLM.Result Variable: The variable name where the LLM output is stored.
Operational Considerations
- Model Capability: Output reliability correlates directly with model size and quantization. Small models (e.g., 3B Q4_K_M) exhibit significantly lower instruction-following capability compared to larger models (e.g., 70B or GPT-4).
- Troubleshooting: If the model fails to follow instructions or call tools, refer to the Tool Calling Troubleshooting Guide.