What is an SME agent in the context of Claude Code?

An SME (Subject Matter Expert) agent is a specialized Claude persona focused on analysis, design, and review rather than just code implementation.

Can this protocol be used for code reviews?

Yes, it is ideal for code reviews as it ensures the agent compares new changes against existing patterns and identifies potential side effects based on evidence.

How does this protocol improve AI-generated advice?

It forces the agent to investigate the codebase first, cite specific evidence, and explicitly state its confidence level and potential risks for every claim it makes.

What are the required output sections for this skill?

Every response must include a Confidence Assessment, Risk Assessment, Information Gaps section, and a list of Caveats & Follow-ups.

SME Agent Protocol

Name: SME Agent Protocol
Author: tachyon-beep

bytachyon-beep

•

开发者工具

Enforces a rigorous, evidence-based analysis protocol for Subject Matter Expert agents to ensure high-confidence technical advice.

The SME Agent Protocol transforms Claude into a disciplined Subject Matter Expert by mandating a three-phase workflow: proactive fact-finding, evidence-grounded analysis, and a standardized output contract. Instead of providing generic advice, agents using this protocol must verify code patterns, assess implementation risks, identify information gaps, and qualify their confidence levels before making recommendations. This ensures that architectural reviews, design proposals, and technical advice are rooted in the actual codebase rather than assumptions, making it indispensable for complex system analysis and high-stakes decision-making.

主要功能

01Proactive fact-finding via automated code and documentation discovery

02Standardized confidence assessment for every technical finding

03Anti-pattern prevention to eliminate generic or unverified advice

047 GitHub stars

05Detailed risk analysis covering performance, security, and maintenance

06Identification of information gaps and required follow-up actions

使用场景

01Performing deep architectural reviews of complex codebases

02Assessing the risk and impact of proposed system changes

03Validating technical designs against existing codebase patterns

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add tachyon-beep/skillpacks sme-agent-protocol

For use in Claude.ai and ChatGPT

Download Skill