prompt-engineering - Design and refine reliable LLM prompts
Design, test, debug, and refine prompts for LLM coding tools, agents, pipelines, personas, rules, skills, and commands.
Tags
Updated: 2026-10-05Capabilities
Typical Inputs
Typical Outputs
What this skill does
- Define measurable prompt objectives
- Draft minimal prompt components
- Build diverse prompt test sets
- Grade prompt outputs
- Diagnose prompt failure causes
- Refine prompts iteratively
- Design autonomous agent personas
- Specify agent tool-use rules
Inputs
- Existing or draft prompt
- Target output requirements
- Representative test inputs
- Evaluation criteria
- Model and version information
- Prompt reference snippets
Outputs
- Refined prompt text
- Prompt test cases
- Evaluation results
- Failure diagnoses
- Agent persona rules
- Prompt design recommendations
Requirements
- Access to an LLM
- An LLM-based tool or pipeline