skills.rest
getsentry
getsentryOFFICIAL
Explore all skills by this author →
Website𝕏@sentryGitHub

prompt-optimizer

Optimize agent prompts across OpenAI, Claude, and Gemini with eval-driven refinements.

905|47|Updated Jan 5, 2026
Search Tags:#workflow#optimization#prompt-engineering#prompt#ai-prompting#eval-loop#model-adapters
One-click install
npx skills add https://github.com/getsentry/skills --skill prompt-optimizer-getsentry
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: prompt-optimizer
Source: https://github.com/getsentry/skills/tree/main/skills/prompt-optimizer
Command: npx skills add https://github.com/getsentry/skills --skill prompt-optimizer-getsentry

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This Skill solves the problem of inconsistent AI prompting by providing an explicit, eval-driven workflow to create, refine, and port prompts and system instructions across model families, ensuring repeatable, auditable results.

Core Features & Use Cases

  • End-to-end prompt optimization workflow: create, refine, port, and evaluate prompts with a structured, critique-driven loop.
  • Model-family adapters: tailor the base prompt for OpenAI, Claude, Gemini, and other families while preserving core behavior.
  • Output artifacts: produces a reusable prompt package with Target, Success Criteria, Optimized Prompt, Adapter Notes, Eval Set, Optimization Log, and Residual Risks for handoff and governance.

Quick Start

Start a first optimization pass on a simple agent prompt and run the meta-optimization loop to iterate toward a validated result.

Frequently Asked Questions about prompt-optimizer

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I optimize AI agent prompts for consistent behavior across OpenAI, Claude, and Gemini?▼

Prompt optimization uses an eval-first loop to iteratively critique and refine system instructions. It produces a reusable package with success criteria, an eval set, and adapter notes for repeatable AI behavior.

What is the best way to evaluate and refine system instructions in a production workflow?▼

Evaluating and refining system instructions requires an explicit, eval-driven workflow that generates an optimization log and residual risks. This structured loop ensures repeatable, auditable AI behavior in production environments.

How do I port a prompt from one model family to another without losing core behavior?▼

To port a prompt across model families, you use model-family adapters that tailor the base prompt while preserving core behavior. The output package includes adapter notes to ensure consistent results across OpenAI, Claude, and Gemini.

Can I use an eval loop to create reusable prompt packages for team handoff?▼

Yes, an eval-first loop generates a reusable prompt package containing the optimized prompt, eval set, and optimization log. This structured output artifact ensures repeatable results and smooth governance during team handoff.

Why does my AI agent prompt produce inconsistent results across different language models?▼

Inconsistent AI prompting happens when prompts lack an explicit, eval-driven workflow. Applying an iterative critique loop with model-family adapters refines system instructions to enforce repeatable behavior across model families.

Related Skills

Explore More Software Engineering Skills→
The-Skyy-Rose-Collection-LLCThe-Skyy-Rose-Collection-LLC
3

eval-harness

Define, run, and report evaluations with /eval commands.

Official
Intermediate
Productfculty-aipmProductfculty-aipm
70

eval-suite-design

Design evaluation suites with three-layer run plans for AI product features.

Community
Advanced
yanochka11yanochka11

prompt-optimizer

Refine prompts by analyzing structure and evaluating variants for clarity.

Community
Advanced
giosuetedeschi-specgiosuetedeschi-spec

prompt-engineering

Design and refine prompts for consistent LLM outputs.

Community
Advanced