chengzhongwei

Local-first prompt coach for Claude Code and Codex that improves prompts, observes prompting habits, and analyzes local history with explicit consent.

11
3
100% credibility
Found Apr 30, 2026 at 11 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
JavaScript
AI Summary

Prompt Sensei is a privacy-focused local coach that analyzes, scores, and improves prompts for AI coding assistants like Claude Code and Codex while tracking personal progress.

How It Works

1
🔍 Discover the coach

While chatting with your AI coding helper, you learn about Prompt Sensei, a friendly guide to make your questions clearer and more effective.

2
📥 Add it to your helper

You download the coach and place it in your AI tool's special folder, then prepare it so it's ready to use.

3
Start your private session

You tell your AI helper to begin coaching, and give permission for it to watch your chats only on your own computer.

4
Improve a rough question

You share a simple question like 'fix this' and instantly get a score, helpful tips, and a polished version that works better.

5
👀 Watch live feedback

As you chat, gentle notes appear showing what's strong and one small habit to try next, like adding more details.

6
📊 Check your progress

You review simple reports on your recent chats to see trends, average scores, and the next skill to practice.

🎉 Prompts that work magic

Your questions now guide the AI perfectly from the start, saving time and making coding fun and frustration-free.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 11 to 11 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is Prompt-sensei?

Prompt-sensei is a local-first JavaScript coach for Claude Code and Codex that observes your prompting habits, improves rough prompts with stage-aware feedback, and analyzes local history only with explicit consent. It turns vague instructions like "fix this test" into structured ones with missing context, verification steps, and one habit to practice next—no cloud, no telemetry. Developers get inline scores, rewrites, and reports right in their AI coding sessions.

Why is it gaining traction?

Unlike cloud-based prompt tools, this local-first GitHub project runs entirely offline, storing only metadata after consent and redacting sensitive data for privacy. The hook is quiet, non-intrusive coaching: start with `/prompt-sensei observe` for real-time tips, use `improve` for quick rewrites, or `lookback` and `report` to spot habit trends across sessions. It focuses on engineering stages like diagnosis or execution, making prompts safer for AI agents that edit code.

Who should use this?

AI coding users relying on Claude Code or Codex in IDEs, especially backend devs debugging tests, frontend engineers implementing features, or anyone iterating on agent workflows. It's for those tired of clarification loops and surprise edits from vague prompts, wanting to build better habits without sharing history externally.

Verdict

Try it if you're deep into Claude or Codex—solid docs and easy skill install make it low-risk, but with 11 stars and 1.0% credibility, it's early-stage and unproven at scale. Pair observed gains with real task outcomes to validate.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.