PermissionGuard

PermissionGuard helps users understand who still controls a token contract, what powers they retain, and why that matters.

It combines AVE risk data with source-based AI analysis to turn contract control into a readable decision.

Why it exists

A token can look liquid, popular, and safe while still giving privileged actors the ability to:

• mint supply
• blacklist wallets
• pause transfers
• upgrade implementation
• change balances or trading behavior

For most users, those are the real trust questions.

PermissionGuard exists to make them answerable in minutes.

What it does

PermissionGuard supports two simple workflows:

1. Search by token name, symbol, or address
2. Analyze a contract directly by address and chain

For each selected contract, it can:

• fetch AVE-backed contract risk signals
• fetch token metadata and holder context
• retrieve verified source code when available
• run AI-based source analysis
• present a readable report focused on permissions, control, and practical user risk

How it works

User
  ↓
Search token or enter contract address
  ↓
Fetch AVE contract + token data
  ↓
Fetch verified source code
  ↓
Run AI source analysis
  ↓
Show a unified permission-risk report

AVE provides the machine-readable first pass. PermissionGuard explains what those signals mean.

Built with

• Next.js
• React
• TypeScript
• AVE API
• Etherscan-compatible source retrieval
• Anthropic-compatible model API
• Vercel Blob caching

Run locally

Requirements

• Node.js 20+
• npm

Environment

Create .env.local:

AVE_API_KEY=***
ETHERSCAN_API_KEY=***
ANTHROPIC_API_KEY=***

# Optional
ANTHROPIC_BASE_URL=...
ANTHROPIC_MODEL=glm-4.7
BLOB_READ_WRITE_TOKEN=...
HTTPS_PROXY=...
HTTP_PROXY=...

Start

npm install
npm run dev

Open http://localhost:3000.

Current scope

PermissionGuard is currently focused on fast, investigation-style contract review.

It works best when:

• AVE data is available for the target token
• verified source code exists on an Etherscan-compatible explorer
• the target contract fits an EVM-style source-analysis workflow

Vision

PermissionGuard is built around a simple idea:

The most useful security product is not the one with the most raw data. It is the one that makes hidden control understandable.

The long-term goal is to become a clear trust interface for onchain products — not just flagging that something may be risky, but showing exactly what power remains after deployment and what that means for users.