openbashok

openbashok / promptzero

Public

Zero trace. Full answer. — Transparent Claude API proxy that anonymizes PII and sensitive data before it leaves your environment. From pentesters, to pentesters.

openbash.com anonymization claude data-privacy infosec llm
24
3
89% credibility
Found May 20, 2026 at 25 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

PromptZero is a privacy-preserving tool that acts as a protective middle layer between you and the Claude AI assistant. When you need help with sensitive information like security findings, client names, IP addresses, or credentials, it automatically detects and replaces this data with safe placeholders before your request reaches the AI. After the AI responds, it restores your real data so you get an accurate, complete answer. This lets security professionals, analysts, and developers safely use AI assistance without risking data exposure.

How It Works

1
🔍 You have sensitive work to do

You need an AI assistant to help analyze security findings, but the data contains real names, IP addresses, and client information you cannot share externally.

2
🛡️ You discover a privacy tool

You find PromptZero, a local tool that acts like a protective wrapper around your AI assistant, keeping your sensitive data safe while still getting helpful answers.

3
⚙️ You set it up once

You install the tool on your computer and connect it to your AI service account. From then on, it runs quietly in the background whenever you use the AI.

4
✨ Your data gets protected automatically

Every time you send a prompt, the tool scans for sensitive information like names, IP addresses, emails, and credentials, replacing them with safe placeholders before your data leaves your computer.

5
🤖 The AI does its job

Your AI assistant receives your question with the protected data, understands what you need, and gives you a complete, thoughtful response.

6
🔄 Your real data is restored

Before the AI's answer reaches you, the tool puts all the real values back in place, so you see the complete, accurate response with your actual data intact.

✅ You get help without exposure

You accomplished your work securely — the AI understood your question, gave you a useful answer, and your sensitive client information never left your computer.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 25 to 24 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 promptzero?

PromptZero is a local proxy that sits between your application and the Claude API, automatically replacing sensitive data with synthetic placeholders before any text leaves your machine. Think of it as a privacy filter for AI prompts: IPs become 198.51.100.x, names become "Soren Brannstrom," hostnames become "alpha.example.com." The LLM processes the sanitized version, you get back a response with real values restored.

Built in Python using spaCy and Presidio for natural language detection, plus regex patterns for network infrastructure. It handles IPv4/IPv6, emails, hostnames, national IDs from Spanish-speaking countries, credit cards, credentials, and more. You point your SDK or CLI to localhost:8000 instead of Anthropic's API and it just works.

Why is it gaining traction?

The pitch is simple and compelling: run your pentest findings through an AI assistant without ever sending real client data to a third-party API. The proxy is transparent to existing code, requires no rewrites. It also includes a Burp Suite integration mode so you can visually verify in your interception proxy that "nexabank" never touches Anthropic's servers.

The documentation is solid for a small project, with clear setup instructions, Docker support, and session management for consistent mappings across multi-turn conversations. Streaming responses are handled correctly, which trips up many similar tools.

Who should use this?

Security professionals and pentesters who want AI assistance writing reports, analyzing findings, or drafting client communications while keeping client data in-scope. Compliance teams operating under NDAs or handling data subject to GDPR will find this valuable too. Anyone running AI tools against sensitive environments where "just don't share that data" is not a realistic workflow constraint.

Verdict

This is a niche tool with a clear purpose and no fluff. The 24-star count reflects its early stage, and the 0.9% credibility score signals you should treat it as a developer-focused project requiring some setup love before production use. Test it thoroughly against your specific data patterns before trusting it with real engagements. For the use case it targets, the approach is sound and the implementation is thoughtful.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.