Purpose: Approved, version-controlled prompt templates for every AI agent/skill, plus evaluation and red-team prompt sets.
What belongs here: System prompts (system/), prompt fragments (skills/), LLM-as-judge evaluation prompts (evaluation/), and adversarial test prompts (red-team/) — all with a version tag and owner.
What must not be stored here: Real candidate/employee data in any example or few-shot content; unversioned or unreviewed prompt changes.
Owner: [TENANT_CONFIGURATION_REQUIRED — AI Governance Lead]
Main dependencies: docs/06-ai-agents-rag/prompt-management.md, src/agents/, evals/.
Subfolders: system/ · skills/ · evaluation/ · red-team/
Status: Initial scaffold — system prompts and red-team test cases populated.