Notes on AI Prompts

Top AI Code Prompt Resources

AI Prompting Tips -- Reddit

  • Structure the prompt in yaml or JSON format.
  • Give the framework first on which to build answer
  • Instead of pros and cons, can ask Top 5 crticisms and advantages.
  • AI is not a mind reader
  • Meta Prompting: e.g. Ask ChatGPT to create good prompt for you.

1. Chatsonic Prompt Library

Free community-driven platform with these key features:

  • 10,000+ code generation prompts across 15+ programming languages[1][2]
  • Curated "Editor’s Choice" section for vetted, high-quality prompts[1-1]
  • Direct integration with Writesonic/Chatsonic API endpoints[1-2]
  • Earn rewards for contributing successful prompts[1-3]

2. AIPRM (AI Prompt Repository)

Enterprise-grade solution for developers:

  • 8,000+ code prompts with version control[1-4][2-1]
  • Team collaboration features for prompt engineering
  • Performance analytics to track prompt effectiveness[2-2]
  • Specialized sections for Python, JavaScript, and cloud infrastructure code[2-3]

4. PromptBase

Marketplace featuring:

  • 15,000+ code prompts with user ratings[1-5][2-4]
  • Advanced filtering by framework (React, TensorFlow) and difficulty level
  • Royalty system for prompt creators[2-5]
  • API access for corporate teams[2-6]

5. FlowGPT

Specialized in ChatGPT code optimization:

  • 6,200+ free coding prompts[1-6]
  • In-browser testing environment for prompts
  • Active community forum for peer reviews
  • Top categories: API integration, algorithm optimization, and DevOps[1-7]

Emerging Tools

Claude Code demonstrates practical implementation through case studies like Simon Willison’s tools.simonwillison.net project, where it helped create Python scripts for commit analysis[3]. ChatX offers a prompt generator with auto-complete features for Midjourney and ChatGPT coding tasks[1-8].

For quality assurance, the CRAAP test (Currency, Relevance, Authority, Accuracy, Purpose) remains essential when evaluating prompts[4]. Leading platforms now display CRAAP scores (50-point scale) for each prompt[4-1].