YCRC has expanded support for AI-assisted coding on its clusters, introducing secure coding environments, locally hosted AI models, and resources for longer-running agentic workflows.
- Claude Code (module load claude): A YCRC-managed, sandboxed environment for research workflows that supports both Yale’s Claude Enterprise service and YCRC’s locally hosted Qwen 3.8 Flash Next model.
- Local Coding Agents (module load local-coding-agents): Provides access to four coding interfaces: Claude Code, Codex, GitHub Copilot CLI, and Pi. They use YCRC’s locally hosted Qwen 3.8 Flash Next (FP8) model with a 262K-token context window. No commercial AI subscription or API key is required. There is no token limit and no associated charges with use.
- Agents partition (salloc -p agent): A new Slurm partition that supports longer-running, lightweight agentic workflows, with jobs lasting up to 7 days. Each job is limited to 1 CPU and 8 GB of memory, with up to two concurrent jobs per researcher.
Both coding environments include YCRC-developed skills that help AI agents interact with cluster software, Slurm jobs, and other research computing resources.
For setup instructions and more information, visit the Coding Agents documentation or contact ycrc@yale.edu.