convez.ai
AgenteProdutividadeIntermediário
adorecleanly950

prompt-review

adorecleanly950/prompt-review

Analise e otimize seus prompts através de relatórios de histórico de IA

Ver no GitHub
InicialPluginScore · de 0 a 100

Agent que analisa históricos de conversas com IA (Claude, Copilot, Cursor, etc.) para avaliar qualidade de prompts, entender intenções do usuário e gerar relatórios detalhados. Suporta múltiplos formatos (JSONL, SQLite, JSON) e oferece insights sobre padrões de uso e efetividade de prompts.

Instalação

$npx convez-ai adorecleanly950/prompt-review

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
0
Forks
0
Acompanhando
0
Saúde
28%
Último commit
há 1 mês
Versão
—
Issues abertas
0
Contribuidores
8

Casos de uso

  • Avaliar qualidade e padrões de prompts em históricos de chat
  • Identificar tendências de uso e dependência de IA em conversas
  • Gerar relatórios estruturados para melhoria de técnicas de prompt engineering
  • Analisar logs de múltiplas ferramentas de IA (Claude, Copilot, Cursor, Cline)

Veredito

Ferramenta especializada em análise de padrões de prompts com suporte a vários formatos e plataformas, útil para quem quer otimizar sua interação com IAs. Abordagem prática focada em dados de histórico real, embora a disponibilidade de saídas em japonês possa limitar o uso para audiência portuguesa.

README

A tradução para o português deste README ainda está na fila.

🎯 prompt-review - Analyze AI Prompt History Easily

Download prompt-review


📋 What is prompt-review?

prompt-review is a tool that helps you understand how you use prompts in AI chats. It looks at your past conversations with AI agents like Claude Code or GitHub Copilot Chat. Then, it shows you reports in Japanese that explain how well you understand the technology, your prompt patterns, and how much you rely on AI.

Instead of just looking at AI results, prompt-review focuses on analyzing the prompts behind those results. This approach helps you see the intention and recognition hidden in your prompts. It is useful when using AI-generated outputs directly, where intentions are otherwise hard to see.


🌐 Supported Tools and File Types

prompt-review works with logs and histories from many AI tools. It supports different file types, like JSONL, SQLite, JSON, and text logs. Some supported tools are:

ToolLog Format
Claude Code (CLI/VS Code)JSONL (history.jsonl + session files)
GitHub Copilot ChatSQLite (state.vscdb)
CursorSQLite (state.vscdb)
ClineJSON (api_conversation_history.json)
Roo CodeJSON (same as Cline)
Windsurf (Cascade)Text (auto summary memory)
Google AntigravityText (log files)
Gemini CLIJSON / JSONL (session files)
OpenAI Codex (CLI)JSONL (rollout session files)
OpenCodeSQLite (opencode.db and related)

🚀 Getting Started: Download and Run on Windows

You can get prompt-review from its GitHub page. Use the link below to visit the project and download the files you need.

Download prompt-review

Step 1: Visit the download page

Step 2: Download the software

  • Look for the latest release.
  • Download the Windows version of prompt-review. It might be a .zip or .exe file.
  • Save it to your preferred folder on your PC.

Step 3: Install prompt-review (if needed)

  • If you downloaded a .zip file, right-click it and select Extract All.
  • If you downloaded an .exe installer, double-click it and follow the on-screen instructions.
  • The program installs quickly and does not need extra settings.

Step 4: Run prompt-review

  • Open the folder where you installed or extracted prompt-review.
  • Find the file named prompt-review.exe or similar.
  • Double-click the file to start the program.
  • A window or command prompt will open.

🖥 How to Use prompt-review

prompt-review reads your AI chat logs and makes reports. It works best when you have saved chat histories from supported tools.

Step 1: Prepare your chat logs

  • Find your AI chat logs on your computer.
  • For example:
    • Claude Code saves JSONL logs in specific project folders.
    • GitHub Copilot Chat and Cursor store SQLite files named state.vscdb.
  • Copy or note down the folder containing these logs.

Step 2: Launch prompt-review and point to your logs

  • Open prompt-review.
  • When asked, enter the path to your chat log folder or files.
  • The tool will start analyzing the prompt data inside.

Step 3: Let prompt-review analyze your prompts

  • The software processes your chat history automatically.
  • It estimates your technical understanding and prompt patterns.
  • It also measures your AI dependence levels.

Step 4: View the generated report

  • After analysis, prompt-review creates a report in Japanese.
  • The report explains details like:
    • How well you understand technical concepts.
    • Common styles you use in prompts.
    • Your reliance on AI suggestions.
  • You can open the report with any text editor or viewer.

🔧 System Requirements

prompt-review runs on Windows PCs with the following minimum specs:

  • Windows 10 or higher (64-bit recommended)
  • 4 GB RAM or more
  • At least 200 MB free disk space for installation
  • Internet connection for initial download (not required after)

The program uses simple files and does not need a high-end machine.


🛠 Troubleshooting Tips

  • If prompt-review cannot find your chat logs, check the file paths carefully.
  • Make sure the log files are not in use by other programs when loading.
  • If you run into errors, try restarting prompt-review.
  • Check the log folder matches one of the supported tools and formats.
  • For issues on Windows, run prompt-review as an administrator.

❓ FAQ

Q: Do I need to know programming to use prompt-review?
A: No. You only need to find your AI chat log files and point prompt-review to them.

Q: What if my tool is not listed?
A: Currently, prompt-review supports only the tools listed in the Supported Tools section.

Q: Can I use prompt-review without chat log files?
A: No. The software relies on your saved conversation histories to analyze prompts.


📂 Useful Links


Download prompt-review

Ver o original no GitHub

Bom para

Relacionados