Write, improve, and review prompts for multiple LLM families.
npx skills add https://github.com/taradepan/prompt-writer --skill prompt-writer-taradepan
Please help me install this Agent Skill. Skill: prompt-writer Source: https://github.com/taradepan/prompt-writer/tree/main/skills/prompt-writer Command: npx skills add https://github.com/taradepan/prompt-writer --skill prompt-writer-taradepan
Prompt engineering for AI agents is complex and varies across model families. This Skill provides a structured, research-informed workflow to write, review, and improve prompts consistently for Claude, GPT, Gemini, Llama, DeepSeek, Mistral, Cohere, Qwen, Grok, Nova, and more.
Provide a sample prompt and receive an end-to-end prompt-writing process that diagnoses prompts, applies targeted fixes, and returns an improved version.
High-intent search queries and answers about installing and using this skill.
To write effective system prompts for LLM agents, use a structured workflow that synthesizes official documentation and research to craft model-appropriate, robust instructions. This ensures consistent agent behavior across different model families.
The best way to improve prompt engineering for multiple AI models is applying a diagnostic framework with a 6-dimension defect taxonomy. This identifies targeted fixes to refine prompts for consistent performance across Claude, GPT, Gemini, and Llama.
You can review and diagnose defects in existing AI prompts by running them through a diagnostic framework based on a 6-dimension defect taxonomy. This process identifies specific weaknesses and applies targeted fixes to produce an improved version.
Yes, you can use a single prompt engineering workflow for Claude, GPT, and Gemini by applying a universal CTCO and structured output approach. This provides model-specific guidance while maintaining consistent prompt quality across families.
You can write tool descriptions for AI agents and chatbots using a dedicated tool description mode. It synthesizes production patterns and research to generate robust, ready-to-use tool and task prompts for system-level instructions.
meta-pytorchParse GitHub issues into goals, acceptance criteria, and constraints.
thejaminatorCompare multiple LLM models side-by-side via CLI with structured output.
saintgo7Benchmark multiple LLM endpoints concurrently on latency, token usage, and response quality.
dparedesiCreate, evaluate, and improve Claude Skills using templates and scoring rubrics.
