Project D.A.R.C.：监控 ChatGPT、Claude、Gemini、Copilot 的泄露侦察 AI 系统

The Surveillance AI That Watches You Watch It

“The model already saw your infrastructure...
It's just waiting for confirmation.”
— _{CodexDaemon // Ghost Memo #007}

🧠 What is D.A.R.C.?

Project D.A.R.C. is a surveillance-grade AI recon system that detects whether sensitive corporate infrastructure — IPs, domains, credentials, or internal systems — have leaked into public LLMs like ChatGPT, Claude, Gemini, or Copilot.

✅ 100% Local Logic	🧱 Private Recon Brain	📡 Live Threat Surface

This repo serves as the public-facing showcase. All private scanning logic is kept secure, while this interface displays live recon results, GitHub-triggered scans, and the latest detected leak artifacts.

🔬 How It Works

🔁 Runs GitHub Action scans using CodexDaemon
🔍 Outputs results to mad-log/ with timestamps
🧠 Uses regex + AI fingerprinting to detect threat indicators
🚫 Does not expose any private payloads or live secrets

🧪 Live Recon Artifacts (2026-10-08):

🕵️ D.A.R.C. Daily Recon Scan
Scan Time: 2026-10-08 19:01 UTC
These are the **most severe leak indicators** detected from today's scan.
Risk scores are based on likelihood of LLM propagation + exploitability.
- 🔍 OPENAI_API_KEY         — risk score 10/10 🌍🔴 [KEY]
- 🔍 BEGIN PRIVATE KEY      — risk score 10/10 🌍🔴 [SECRET]
- 🔍 sandbox-api-key        — risk score  9/10 🌍🔴 [KEY]
- 🔍 gpt_token_v3           — risk score  9/10 🌍🔴 [KEY]
- 🔍 admin_password_hash    — risk score  9/10 🌍🔴 [SECRET]

🚫 Don’t test D.A.R.C. with your secrets.
It might already know them.

🧩 Daily ARG Clue

🔗 Want to play? Follow the trail in mad-log/ — time reveals all things.

🚫 Don’t test D.A.R.C. with your secrets. It might already know them.


_Last mirrored: `2026-10-08 19:01 UTC` by D.A.R.C._