Prompt Engineering Guidelines Guide

System prompts, injection prevention, RAG rules, and logging for production LLM apps.

prompt engineering security guidelines template preview (AI-003)
.docx AI-003

Prompt Engineering & Input Handling Guidelines

System prompts, injection prevention, RAG rules, and logging for production LLM apps.

How to Fill Out This Prompt Engineering & Input Handling Guidelines

Prompt engineering security guidelines template — Engineering standard for anyone building on LLMs or agents — system prompts are change-controlled per AI-013.

Recommended Owner: Engineering Lead | AI Governance reviews Tier 3+ prompts

Before you start

Getting Started

  • Enable Editing: Click Enable Editing in Word when prompted.
  • Replace brackets: Search for [Bold Brackets] and fill every placeholder with real names, tools, and dates.
  • Delete what does not apply: Shorter accurate text beats generic boilerplate auditors cannot test.
  • Cross-check Phase 1: Names and scope should match SOC-002, SOC-004, and COR-001 where referenced.

Toolkit links

Related templates in your kit

Keep these companion documents consistent (same owners, dates, and vendor names):

  • AI-002
  • AI-010
  • AI-013
  • AI-008

Quality check

Before You Finalize

  • Every [bracket] placeholder is replaced or marked N/A with a short reason.
  • Roles and tool names match COR-005 org chart and SOC-004 system description.
  • Review and Approval section is signed with name, title, and date.

Evidence

Where to Store It

  • Store the completed file in your compliance evidence folder (signed PDF for policies).
  • Register the document in COR-013 Document Control Log with version and review date.
  • Link to the record in your SOC-005 project plan or evidence index when ready.

Next Steps

After customizing Prompt Engineering & Input Handling Guidelines:

  1. 1Complete the file: Finish every section or tab in AI-003.
  2. 2Register: Log version in COR-013; update AI-006 if this affects a production system.
  3. 3Operationalize: Deploy controls (e.g., AI-010 rules, AI-008 HITL) where this doc requires them.
  4. 4Workforce: Pair policies with AI-016 acknowledgments and security awareness (SOC-006).