chandika

chandika / mirage-proxy

Public

AI now sees plausible fakes of API keys, PII, credentials. Rust. Sub-millisecond.

18
0
100% credibility
Found Feb 22, 2026 at 11 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Rust
AI Summary

Mirage Proxy is a lightweight local tool that intercepts messages to AI services, swaps sensitive personal details and secrets with realistic fakes, and swaps them back in replies to prevent leaks.

How It Works

1
🔍 Discover the protector

You hear about Mirage Proxy when you're worried your AI helpers might accidentally see real passwords or emails from your files.

2
📥 Get it set up

You download and install the protector easily, choosing to build it yourself for trust or using a quick helper tool.

3
🛡️ Turn on the shield

With one command, you activate the background protector that watches all your AI chats and keeps real info safe.

4
💬 Chat with AI tools

Open your favorite AI coding helper or chat app, and everything flows normally but safely through the protector.

5
📊 Watch it catch secrets

You glance at the live log and smile seeing fake emails and keys replacing the real ones invisibly.

6
🔧 Toggle anytime

Use simple words like 'on' or 'off' to control it per chat window, or fully remove if needed.

✅ AI magic without worry

Now you brainstorm code and ideas freely, knowing your private info stays locked on your machine forever.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 11 to 18 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 mirage-proxy?

Mirage-proxy runs as a local daemon proxy between your LLM tools and APIs like Anthropic or OpenAI, scanning requests for secrets, GitHub tokens, emails, phones, and PII, then swapping them with format-matching fakes—like turning AKIA... into another valid-looking AWS key. Real data never leaves your machine; responses get originals restored via session-consistent mappings. Built in Rust for sub-millisecond latency, it auto-routes 28+ providers and installs via cargo or brew with shell integration for tools like Cursor, Aider, or Claude Code.

Why is it gaining traction?

Unlike tools that insert obvious [REDACTED] tokens—prompting models to complain or adapt—mirage-proxy's invisible sensitive replacements keep LLMs oblivious, preserving natural behavior. Zero-config service install sets env vars like ANTHROPIC_BASE_URL to localhost:8686, supports SSE streaming, and offers dry-run mode plus an encrypted vault for persistent mappings. At ~5MB with <1ms overhead, it crushes bloated alternatives like LLM Guard or LiteLLM+Presidio in speed and size.

Who should use this?

Backend devs building AI agents that ingest repo context full of keys, or Cursor/Aider users tired of leaked GitHub tokens in prompts. Anyone routing codebase prompts to cloud LLMs via OpenClaw, Continue, or raw OpenAI calls, especially if sandboxing falls short on network-layer leaks.

Verdict

Grab it if you're proxying LLM traffic with secrets in play—docs are solid, MIT-licensed, and service rollback is one command away. With 11 stars and 1.0% credibility, it's early alpha; test dry-run first before production.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.