23. Agents on the Cluster#
This chapter covers running agentic AI tools on the Kempner AI cluster:
Using Agentic AI on the Cluster: install and sign in, permission modes, keeping a session alive, IDE agents, and responsible use on shared infrastructure.
Your First Agentic Workflow on the Cluster: one small task end to end, from launching the agent to checking the result.
Agent Security and Scoping: prompt injection, remote control, and how to limit what an agent can reach, including the agent sandbox.
HPC Agentic Recipes: serving open-weight models on the cluster’s GPUs, so your prompts and code never leave the cluster.
To give an agent your project’s conventions and guardrails, continue to Configuring and Extending Agents.