What does the Prompt Audit skill actually analyze?

It reviews your stored session journals within the ctx system to identify turns where the AI had to ask clarifying questions instead of performing the requested task.

How many sessions do I need for an audit?

You need at least 3 recent sessions stored in your journal. This ensures the skill has a large enough sample size to identify genuine patterns rather than one-off mistakes.

Will this skill provide feedback automatically?

No, it is designed to be user-invoked. It only runs when you explicitly call the command, ensuring that coaching feedback is provided only when you are in a mindset to receive it.

Does it criticize every short prompt?

No, the skill is context-aware. It recognizes when shorthand is efficient (e.g., following an established pattern) versus when it is truly vague and stalls the workflow.

How does it help me improve my prompts?

It provides a 'Prompt Audit Report' that quotes your specific messages and offers a concrete 'Better Prompt' alternative for each example, making the advice actionable.

Prompt Audit & Optimization

Name: Prompt Audit & Optimization
Author: ActiveMemory

byActiveMemory

•

生产力与工作流

Analyzes session history to provide actionable feedback on prompting patterns and reduce AI clarification cycles.

The ctx-prompt-audit skill is a performance-tuning tool for Claude Code designed to optimize the communication loop between developers and AI. By reviewing historical session transcripts stored in the ctx journal, it identifies specific instances where vague instructions led to unnecessary back-and-forth, such as missing file context or ambiguous targets. The skill generates a comprehensive coaching report that includes concrete examples of your actual prompts, suggested rewrites, and an analysis of your recurring habits, helping you achieve higher precision and efficiency in your AI-assisted development workflow.

主要功能

01Automated transcript analysis of recent coding sessions

02Context-aware differentiation between efficient shorthand and ambiguity

03Pattern recognition to track both strengths and recurring weaknesses

0436 GitHub stars

05Personalized coaching reports with concrete 'Better Prompt' rewrites

06Identification of vague prompts and missing technical context

使用场景

01Troubleshooting high rates of clarification cycles during complex refactors

02Onboarding developers to more effective AI-assisted coding habits

03Refining AI communication style to reduce token usage and time-to-solution

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add activememory/ctx ctx-prompt-audit

For use in Claude.ai and ChatGPT

Download Skill