Launching today

anubis-guardian
Deterministic repository security for AI-assisted code
2 followers
Deterministic repository security for AI-assisted code
2 followers
Open-source, multi-layer repository security scanner for AI-built projects. Detects secrets, Git-history leaks, CI/CD and infrastructure misconfigurations, dependency risks, and selected JS/TS code...

Free
Launch tags:GitHub
Launch Team

AWS Startup AdvisorPersonalized AWS guidance that meets you where you’re already building
Promoted
Hey Product Hunt community! ⚱️
As codebases are increasingly built by AI assistants and LLMs at unprecedented speeds, we are encountering a massive, quiet crisis: Vibe Coding Vulnerabilities. AI agents write beautiful code, but they consistently leave behind critical logic flaws, loose credentials, unprotected database Row-Level Security (RLS), and dangerous client-server exposure.
That’s why I built Anubis Guardian.
It started as the core preventative defense engine for a massive browser-based enterprise system I’m building (MY ARENA OS). I realized that every single software creator leveraging AI tools needs this exact layer of immune defense right now. So, I open-sourced it today.
What makes Anubis Guardian fundamentally different?
Zero Runtime Dependencies: Built strictly using native Node.js and commander. It is completely immune to third-party supply-chain attacks.
Hard Honesty Contracts & Masking: It masks every single exposed credential using SHA-256 before any AI-review token is dispatched. Your data stays yours.
Deterministic Verification: It features a unique verifier layer that completely eliminates LLM hallucinations. If an AI hunter suggests a bug, it must be proved against a deterministic structural match to be confirmed.
Preventative Guard Hook: It sits directly inside the loop at write-time, blocking RLS-killing SQL and secret literals before they ever get committed.
Anubis Guardian is completely free, open-source, and offline-first. Check out our live repo, run npm install -g anubis-guardian in your workspace, and let me know your brutal, honest feedback!
Excited to chat with you all in the comments!
3. Product Description (Short & Punchy)
Anubis Guardian is an offline-first, zero-dependency repository security scanner specifically optimized for AI-assisted workflows and modern tech stacks (Next.js, Supabase, NestJS). It bridges the gap between raw secret scanners and heavy AST engines by utilizing a dual-agent mechanism: an AI hunting engine backed by a strict, independent deterministic verifier to kill hallucinations.
With its signature write-time prevention hooks (init-hook), it stops AI agents from committing insecure patterns—like client-exposed service roles or world-writable database policies—before they ship to production.
4. Key Features & Bullet Points (For the PH Listing)
hall-of-fame-level Immunoassay: Zero third-party runtime dependencies to block external exploit injection.
Dual-Agent Verification Architecture: Isolated hunters spot vulnerabilities; a strict deterministic verifier independently approves or rejects to eliminate false positives.
Universal Repository Cover: Scans credentials in configuration formats, committed .env leaks, and hidden traces in Git history across any language stack.
Deep Native JS/TS & Framework Stacks: Deep code analysis for Next.js App Router boundary errors, Supabase RLS omissions, and data-leakage query shapes.
Preventative Guard Hook: Optional write-time guard that safely intercepts and blocks high-risk agent generation at execution point.
Consent-Gated URL Audit (Mode 2): Secure live probing that handles SSRF prevention, IP pinning, and verified ownership barriers before active testing.
5. First 3 FAQs (Frequently Asked Questions for the Launch Page)
Q1: How does this overlap or differ from Gitleaks or Semgrep?
Gitleaks focuses solely on secret pattern shapes. Semgrep provides broad static analysis based on curated rulesets but lacks a strict scoring contract or zero-config out-of-the-box specialization for modern AI-stack footprints. Anubis Guardian merges universal surface scanning (history, Docker, CI pipelines) with unique application logic scanning (tenant isolation, authorization evidence) tailored specifically to counteract AI coding agent footguns.
Q2: Is it safe to feed my code into the AI review feature?
Yes. Anubis Guardian operates under a strict offline-first protocol. Even when you explicitly call the --ai-review layer, all credentials, API keys, and sensitive tokens are completely masked into cryptographically safe SHA-256 signatures before being parsed. Your live production secrets never leave your local environment.
Q3: What is the relation to MY ARENA OS?
Anubis Guardian was engineered as the foundational native security module that safely guards the development ecosystem of MY ARENA OS—an all-in-one cloud operating system. Due to its sheer utility for standalone developers using cursor, Claude Code, or Copilot, the scanning engine was decoupled and made available as a sovereign open-source tool for the broader developer community.