Prompt Optimizer
Optimize prompts for Claude 4.x models (Sonnet 4.5, Haiku 4.5, Opus 4.5) using Anthropic's official guidance.
Core Principles
- Be Explicit
Claude 4.x models follow instructions precisely. Request behaviors explicitly rather than hoping Claude infers them.
Less effective: Create an analytics dashboard
More effective: Create an analytics dashboard. Include as many relevant features and interactions as possible. Go beyond the basics to create a fully-featured implementation.
- Provide Context for Why
Explain motivation behind instructions to help Claude generalize appropriately.
Less effective: NEVER use ellipses
More effective: Your response will be read aloud by a text-to-speech engine, so never use ellipses since the text-to-speech engine will not know how to pronounce them.
- Align Examples with Desired Behavior
Claude pays close attention to examples. Ensure they demonstrate exactly what you want, without patterns you want to avoid.
Optimization Workflow
-
Diagnose the issue (verbosity, formatting, tool use, instruction following)
-
Select relevant patterns from references/patterns.md
-
Apply the appropriate prompt snippets
-
Test the optimized prompt
Quick Fixes by Issue
Too Verbose
Add: Keep responses concise and direct. Avoid unnecessary elaboration.
Too Much Markdown/Bullets
See references/formatting.md for detailed prompts to control output style.
Not Using Tools
Add: By default, implement changes rather than only suggesting them. Use tools to take action.
Generic "AI Slop" Output
Add creativity guidance or use the frontend-aesthetics pattern from references/patterns.md.
Not Following Instructions
Make instructions explicit, add context for why, frame positively (what TO do, not what NOT to do).
Reference Files
-
references/patterns.md - Complete prompt snippets organized by use case
-
references/formatting.md - Output format control patterns
-
references/agentic.md - Tool use, parallel execution, and multi-context workflows
Load the appropriate reference file based on the optimization need.