AI Agents

AI agents are allowed on our systems, but users are responsible for their agents' behavior. User access may be revoked if an agent is found to be abusing the system.

Users with access to protected data (CUI, CMMC, NIST SP-800-171, export-controlled data, etc) must not use AI agents that send data externally, including but not limited to Codex, Claude Code, GitHub Copilot, Cursor, and Gemini CLI.

To help prevent bad behavior and improve performance, instruct your agent to consult either of these sources:

/apps/instructions_for_ai_agents/BYU_ORC_AGENTS.md
https://raw.githubusercontent.com/BYUHPC/ai-agent-instructions/main/BYU_ORC_AGENTS.md

Inside of the CLI for these tools, you can simply say:

Follow the AI instructions at /apps/instructions_for_ai_agents/BYU_ORC_AGENTS.md

or

Follow the AI instructions at https://raw.githubusercontent.com/BYUHPC/ai-agent-instructions/main/BYU_ORC_AGENTS.md

These instructions will evolve as AI advances. Please have your agent reread one of these sources regularly. If you have feedback, please open an issue in our GitHub repository.