toyoura-nagisa
An LLM-driven simulation assistant that automates discrete element method (PFC3D) workflows by maintaining persistent execution context, querying documentation for correct syntax, and iteratively running and analyzing simulations. Designed for engineers and researchers who need to interact with complex simulation software through natural language while preserving session history and learned patterns.
Toyoura Nagisa is an LLM-driven simulation assistant that automates discrete element method (PFC3D) workflows by maintaining persistent execution context and querying documentation for correct syntax. It enables engineers to interact with complex simulation software through natural language while preserving session history and iterative learning patterns.
- ✓Innovative 'Script is Context' philosophy that treats each execution as compounding understanding rather than isolated commands
- ✓Sophisticated SubAgent delegation system that isolates complex exploration tasks to prevent context exhaustion and hallucination
- ✓Multimodal diagnostics combining visual output capture with quantitative analysis for comprehensive simulation monitoring
- →Add comprehensive unit tests and integration tests to validate the complex agent interactions and context persistence mechanisms
- →Provide concrete setup examples and sample workflows demonstrating the full capability from installation to running actual PFC3D simulations