AI & agents
A local workbench for finding the right model, gateway, and agent setup for the job.
Behind the build
- The problem
- Comparing AI gateways means juggling configurations, responses, costs, and decisions across disconnected tools.
- The approach
- Built a shared playground for model comparisons, benchmarks, MCP exploration, and Docker-based coding-agent sandboxes.
- What it enables
- Prompts, run history, and model decisions stay together in a local workspace, with agent experiments isolated from everyday CLI settings.



