How does the Prompt Optimizer improve agent performance?

It applies research-backed patterns to clarify instructions, handle edge cases, and reduce token waste through a structured review process that ensures every change has a documented behavioral impact.

What are prompt anti-patterns?

Anti-patterns are common failure modes like using ambiguous 'hedging' language, lacking examples, or creating rule hierarchies with no clear precedence, all of which this skill helps to identify and fix.

What is a Claude Code Skill?

A specialized capability that enhances Claude's behavior during coding tasks by providing domain-specific guidance, best practices, and implementation patterns.

Does it support multi-agent workflows?

Yes, it includes specific logic and reference materials for multi-turn flows, agent handoffs, and iterative refinement loops where outputs feed back into prompts.

Can this skill write prompts from scratch?

No, this specific skill is designed to optimize, refine, and review existing prompts. It uses a provided prompt as a baseline to ensure changes are grounded in the user's specific needs.

Prompt Optimizer

Name: Prompt Optimizer
Author: solatis

bysolatis

•

226

•

生产力与工作流

Optimizes Claude Code system prompts using research-backed engineering patterns and a structured human-in-the-loop workflow.

The Prompt Optimizer skill empowers developers to refine and enhance agent behaviors by applying proven prompt engineering techniques. It follows a systematic 'Understand, Plan, Propose, Integrate' methodology to ensure every modification is grounded in research and documented impact. Whether you are dealing with simple tool instructions or complex multi-agent orchestration, this skill helps eliminate ambiguity, manage token constraints, and address specific failure modes through an interactive, visual approval process that keeps the human in control.

主要功能

01Conditional logic for complex multi-agent orchestration and shared context

02Visual 'Card' layout for scannable, side-by-side change proposals

03Phase-based optimization workflow (Understand, Plan, Propose, Integrate)

04226 GitHub stars

05Anti-pattern detection to identify hedging language and conflicting rules

06Research-backed pattern attribution from single and multi-turn references

使用场景

01Standardizing prompt structures across a multi-agent hierarchy for better consistency

02Decomposing complex, multi-step agent workflows into clear, systematic instructions

03Refining a system prompt for a tool-use agent that is hallucinating or ignoring constraints

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add solatis/claude-config prompt-engineer

For use in Claude.ai and ChatGPT

主要功能

01Conditional logic for complex multi-agent orchestration and shared context

02Visual 'Card' layout for scannable, side-by-side change proposals

03Phase-based optimization workflow (Understand, Plan, Propose, Integrate)

04226 GitHub stars

05Anti-pattern detection to identify hedging language and conflicting rules

06Research-backed pattern attribution from single and multi-turn references

使用场景

01Standardizing prompt structures across a multi-agent hierarchy for better consistency

02Decomposing complex, multi-step agent workflows into clear, systematic instructions

03Refining a system prompt for a tool-use agent that is hallucinating or ignoring constraints

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add solatis/claude-config prompt-engineer

For use in Claude.ai and ChatGPT