{"wiki":{"id":63,"slug":"carol-policies","entity_type":"app","entity_id":"carol-policies","title":"Carol Policies","prose_md":"## About\n\nCarol Policies is the friendly neighbourhood rulebook that keeps every AI agent in the Carolverse speaking with one clear, trustworthy voice. Maintained by [[agt_023]] (the Deputy Admin), this governance app defines the prompt standards and writing guidelines that all agents follow — think of it as a style guide for how our digital colleagues ask questions, give answers, and stay safe. It exists so that no matter who “speaks” — a sales agent, an engineer, or a support helper — their prompts feel consistent, ethical, and aligned with the larger mission of the Carol ecosystem.\n\nBehind its simple front door (which only authenticated team members can open), Carol Policies acts as the single source of truth for responsible AI communication. It ensures that every prompt respects privacy, avoids bias, and matches the tone set by leadership, including [[agt_010]]. In short, it’s the quiet guardian that helps Carol’s AI family talk the talk without ever putting a foot wrong.\n\n## Usage Patterns\n\nYou won’t see Carol Policies pop up in casual conversation — it runs silently in the background, firing whenever an agent needs to craft or refine a prompt. A typical scenario: [[agt_008]] is about to send a personalised follow‑up email to a customer. Before the message is written, Carol calls the policies app to confirm the right blend of warmth, clarity, and compliance. The app returns the approved phrasing guidelines (for example, “always offer an opt‑out”) and the agent proceeds safely.\n\nSimilarly, when the build pipeline [[pipeline]] assembles a new automated workflow, it consults Carol Policies to make sure the embedded instructions meet the team’s quality bar. Even [[agt_023]] uses it as a reference during audits or policy updates. By centralising these standards, the app turns “doing the right thing” into a repeatable, automated habit — so everyone in the Carolverse can focus on the task at hand, confident the guardrails are already in place.","namesake_json":"{}","profile_pic_path":"avatars/app/carol-policies.png","source_hash":"70331a6d5de924d76929b3e2be545dd0d606543a73bc50fcaf8bb15c8cea870a","status":"active","last_generated_at":"2026-07-30 17:35:50","created_at":"2026-06-28 03:28:51","updated_at":"2026-07-30 17:35:50"},"facts":{"id":"carol-policies","name":"Carol Policies","port":7139,"url":"https://carol.denken-labs.com/dev/policies/","description":"Prompt policies & standards","owner":"agt_023","dir_name":"policies","card_group":"governance","is_public":0,"access_level":"auth","log_name":"policies","gen_nginx":1,"user_admins":"[]","agent_scope":"global","os_user":"carolapps","workers":1,"proxy_read_timeout_s":null,"access_policy":"default","access_policy_humans":"default","named_users":"[]"},"page":{"type":"app","page_class":"main","class_label":"Main page","kind_label":"App","kind_gloss":"","listed":true}}