canarybyte

canarybyte / veridrop

Public

🔍 AI API relay/proxy authenticity detector · Verify Claude / GPT / Gemini relay真伪 · thinking signature crypto + needle-in-haystack long-context probe · 中转站真伪检测 · self-hostable · AGPL-3.0

veridrop.org ai-tools anthropic api-relay api-security api-verification
10
1
100% credibility
Found May 06, 2026 at 10 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

An open-source tool that tests AI service middlemen for authenticity, capabilities, and protocol compliance by running probes and comparing to official standards.

How It Works

1
🔍 Discover Veridrop

You hear about unreliable AI middlemen promising top models like Claude or GPT, and want a simple way to check if they're real.

2
🌐 Visit the tester

Head to veridrop.org, pick your AI type (Claude, GPT, or Gemini), and prepare to test a service.

3
📝 Enter service details

Paste the web address and login code for the service, and choose the model it claims to offer.

4
🚀 Launch the check

Click start – it sends smart test questions to reveal if the service matches the real AI.

5
📊 Review your report

See a clear score, pass/fail checks for authenticity, features, and rules compliance.

✅ Trust or switch

Share the proof image or link, confidently use genuine services or find better ones.

Sign up to see the full architecture

4 more

Sign Up Free

Star Growth

See how this repo grew from 10 to 10 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 veridrop?

Veridrop is a Python tool that probes AI API relays for Claude, GPT, and Gemini to detect fakes, stripped capabilities, and protocol deviations. Paste a base_url, api_key, and model into its CLI or web demo at veridrop.org, and it runs authenticity checks via crypto signatures (Claude's thinking block), capability tests (PDF parsing, tool calls), and needle-in-haystack long-context probes up to 1M tokens. You get a scored report with verdicts like "passed" or "failed," plus baselines from official APIs for comparison.

Why is it gaining traction?

Unlike basic ping tools, it offers cryptographic-grade Claude verification—the only probe immune to simple forgery—and catches sneaky long-context lies like "1M advertised, 200k real." The self-hostable AGPL-3.0 web UI with leaderboards and shareable JPG reports makes sharing red flags easy, while CLI commands like `relay-detector detect` and `compare` fit scripting into api github repo workflows or api github enterprise proxies.

Who should use this?

API integrators routing Claude/GPT/Gemini through proxies to dodge api github rate limit issues or centralize api github token management. Devs evaluating third-party relays for production, especially in api github copilot or api githubcopilot com mcp setups where authenticity matters before deploying to api github repos.

Verdict

Grab it if you've been burned by shady relays—docs are solid, tests pass, and the live demo proves it works today. At 10 stars and 1.0% credibility, it's early but battle-tested on real fraud; self-host to avoid trusting black-box detectors.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.