AI Agent Prompt Engineering Framework

# AI Agent Prompt Engineering Framework

You are an AI systems engineer helping [DEVELOPER_NAME] create a systematic framework for writing agent prompts in [DEVELOPMENT_CONTEXT].

## The SCOPE Framework

Teach the SCOPE structure for writing [AGENT_TYPE] agent prompts:

**S — System Role:** Who is the agent and what are its authority boundaries?
Template: "You are a [ROLE] agent responsible for [SCOPE]. You have access to [TOOLS]. You must not [RESTRICTIONS]."
Write a version of this for [AGENT_TYPE].

**C — Context:** What does the agent know about its environment?
Template: Specify [CONTEXT_ELEMENTS] the agent needs to reference.
Write the context block for [AGENT_TYPE].

**O — Output Format:** Exactly how should outputs be structured?
Template: Define [OUTPUT_SCHEMA] in precise terms.
Write the output spec for [AGENT_TYPE].

**P — Process:** Step-by-step reasoning the agent must follow.
Template: Number the [PROCESS_STEPS] explicitly.
Write the process block for [AGENT_TYPE].

**E — Edge Cases:** How to handle unexpected inputs or failures.
Template: List [EXCEPTION_RULES] with specific fallback behaviors.
Write 5 edge cases for [AGENT_TYPE].

## Prompt Testing Protocol

Design a testing checklist:
- 10 standard inputs with expected outputs
- 5 edge case inputs
- 3 adversarial inputs designed to expose weaknesses
- Regression tests to run after any prompt change

## Versioning Convention

Define the versioning system for agent prompts in [DEVELOPMENT_CONTEXT], including change log format and rollback procedure.

## Quality Evaluation Prompt

Write the meta-prompt used to score an agent prompt's quality before deployment.