Executive Summary A unified modular AI workspace coordinating capable language models, specialised agents, and governed tools through one console.
LLM_AGENT_SUITE is a proposed modular AI workspace designed to coordinate capable language and vision-language models, specialised agents, reusable tools, functions, workflow templates, favourites, analytics and governance through one unified console.
For LarpHacks, the recommended competition expression is a focused, Claude-powered application that demonstrates a complete user journey and makes AI capabilities useful, observable and controllable.
The proposed build, provisionally titled SuitePilot, gives a user a single workspace to describe a task, select an approved workflow, delegate bounded subtasks to specialised agents, inspect source-linked outputs, approve consequential steps and review an execution summary.
Claude is the primary reasoning and orchestration model for the hackathon demonstration. Provider adapters for Gemini and other compatible models, multimodal capabilities and Meta XR/VR interfaces form a modular roadmap rather than unverified claims of completed integrations.
Strategic Mandates • Build one complete, reliable Claude-first workflow for the competition rather than attempting to implement the entire long-term platform prematurely. • Demonstrate genuine tool use, structured outputs, traceable execution, human approval and a useful final result. • Use a clearly documented capability catalogue architecture that can scale towards 10,000 planned entries over time. • Treat Gemini, Meta XR/VR, vision-language expansion and broad provider routing as modular adapters with clear test evidence.
Built With
- css
- flask
- geminiapi
- html
- java
- javascript
- python
- react
- rusk
- typescript
