When is the Prompt Refinement skill automatically invoked?

This skill is triggered at the start of major ai-eng commands, including /ai-eng/research, /ai-eng/specify, /ai-eng/plan, and /ai-eng/work.

How does the confidence assessment work?

It evaluates the refined prompt on a scale from 0.0 to 1.0. A score of 0.8-1.0 indicates all ambiguities are resolved, while lower scores highlight areas where assumptions were made.

Can I manually edit the refined prompt?

Yes. After the refinement process, you are presented with a confirmation step where you can choose to proceed, ask more questions, or manually edit the resulting prompt.

What is the TCRO framework?

TCRO stands for Task, Context, Requirements, and Output. It is a systematic framework used to provide clear, actionable, and structured instructions to an AI.

Prompt Refinement

Name: Prompt Refinement
Author: v1truv1us

byv1truv1us

•

Productivity & Workflow

Transforms vague user inputs into structured, high-quality prompts using the TCRO framework to maximize AI performance and consistency.

About

The Prompt Refinement skill is a core utility designed to eliminate ambiguity in AI-assisted workflows by converting messy input into structured specifications. By applying the TCRO (Task, Context, Requirements, Output) framework, it ensures that every command—from research to implementation—starts with a clear, actionable directive. This skill reduces iteration cycles and improves output consistency by asking targeted, phase-specific clarifying questions and applying advanced incentive prompting techniques like expert persona assignment, stakes language, and step-by-step reasoning, while integrating project-specific context from CLAUDE.md.

Key Features

Confidence Assessment Scoring (0.0 to 1.0 scale)
Automated Incentive Prompting and Persona Assignment
2 GitHub stars
Phase-Specific Clarification (Research, Specify, Plan, Work)
Project Context Integration via CLAUDE.md analysis
TCRO Framework Structuring (Task, Context, Requirements, Output)

Use Cases

Refining research tasks to ensure targeted and actionable data collection.
Converting one-line feature requests into comprehensive technical specifications.
Standardizing development workflows to align with existing project architectural patterns.

About

Key Features

Confidence Assessment Scoring (0.0 to 1.0 scale)
Automated Incentive Prompting and Persona Assignment
2 GitHub stars
Phase-Specific Clarification (Research, Specify, Plan, Work)
Project Context Integration via CLAUDE.md analysis
TCRO Framework Structuring (Task, Context, Requirements, Output)

Use Cases

Refining research tasks to ensure targeted and actionable data collection.
Converting one-line feature requests into comprehensive technical specifications.
Standardizing development workflows to align with existing project architectural patterns.