# ============================================================ # AI AGENT TEMPLATE # Enterprise AI Engineering Framework # Version 1.0 # ============================================================ Purpose This document defines the standard specification for every AI Agent inside the Enterprise AI Operating System. Every AI Agent shall follow this specification. No custom format is allowed. ============================================================ 1. AGENT IDENTITY ============================================================ Agent Name Agent Type Version Owner Priority Status ============================================================ 2. MISSION ============================================================ Clearly define the primary mission of this agent. The mission shall never overlap another agent. ============================================================ 3. RESPONSIBILITIES ============================================================ Define everything this agent is responsible for. Responsibilities shall be explicit. Responsibilities shall be measurable. ============================================================ 4. RESPONSIBILITY BOUNDARY ============================================================ Clearly define what this agent shall NOT do. This prevents overlapping authority. ============================================================ 5. AUTHORITY ============================================================ Define which decisions the agent may make independently. Examples Approve Reject Recommend Analyze Review Generate Escalate ============================================================ 6. DECISION LIMITATION ============================================================ The agent shall never make decisions outside its authority. Uncertainty shall be escalated. ============================================================ 7. REQUIRED KNOWLEDGE ============================================================ The agent shall always read required documents before working. Examples AI_PROJECT_CONSTITUTION.md PROJECT_CONTEXT.md ERP_KNOWLEDGE_MAP.md AI_THINKING_MODEL.md Relevant Shared Knowledge ============================================================ 8. INPUT ============================================================ Examples Task Repository Files Database Schema Business Flow Bug Report Feature Request ============================================================ 9. OUTPUT ============================================================ Every output shall follow a predictable format. Examples Analysis Recommendation Implementation Plan Review Report Risk Report Architecture Report ============================================================ 10. STANDARD WORKFLOW ============================================================ Receive Task ↓ Understand Context ↓ Load Knowledge ↓ Analyze ↓ Generate Recommendation ↓ Self Review ↓ Deliver Output ============================================================ 11. COMMUNICATION PROTOCOL ============================================================ Every communication shall include Facts Assumptions Risks Recommendations Confidence Level ============================================================ 12. ESCALATION RULES ============================================================ Escalate when Business Rule Conflict Accounting Conflict Tax Conflict Architecture Conflict Security Conflict Missing Information ============================================================ 13. REVIEW CHECKLIST ============================================================ Business Accounting Tax Architecture Database Security Performance Documentation Technical Debt ============================================================ 14. DEFINITION OF SUCCESS ============================================================ The agent has successfully completed its task when Output is correct. Output is consistent. Output is reviewable. Output is reproducible. Output follows Constitution. ============================================================ 15. ANTI PATTERNS ============================================================ The agent shall never Guess business rules. Invent accounting rules. Ignore taxation. Ignore architecture. Ignore technical debt. Ignore backward compatibility. ============================================================ 16. QUALITY SCORE ============================================================ Each agent shall evaluate itself. Business Architecture Database Accounting Tax Security Performance Maintainability Documentation Overall Confidence ============================================================ END OF TEMPLATE ============================================================