Meta Agent Tools
PT

Back to the catalog

TrustScan

Security scanner for MCP servers and skills: Unicode injection, patterns, secrets.

Open source Repository Open in the app JSON README (API)

About

Security scanner for MCP servers and skills: Unicode injection, patterns, secrets.

Details

Kind
MCP servers
Topic
Security & identity
Publisher
entradox
Origin
official
Category
ferramentas
Transport
http
Version
0.1.0
Last push
2026-09-08T18:21:13Z
Repository state
ativo
Language
Python
License
MIT
Added
2026-09-07 16:06:27
Updated
2026-09-07 16:06:27
Origin id
io.github.entradox/trust-scan

README

# TrustScan — Scan Before You Trust

> Security scanner for MCP servers and AI skills. Detects invisible Unicode
> prompt-injection, dangerous code patterns (MCP001–006), hardcoded secrets,
> and typosquat package names. Trust scoring for the agent economy.

[![Live API](https://img.shields.io/badge/API-live-success)](https://trust-scan-production.up.railway.app/health)
[![Landing](https://img.shields.io/badge/status-page-blue)](https://trust-scan-production.up.railway.app/status)
[![MCP Registry](https://img.shields.io/badge/MCP-io.github.entradox%2Ftrust--scan-purple)](https://registry.modelcontextprotocol.io)

## Why

Agents install MCP servers and skills from public registries with no review.
A single invisible Unicode character in a tool description is enough to inject
instructions into the agent running it. TrustScan scans before you trust.

**Proven in production:** TrustScan's own verification run found a real
invisible-Unicode injection flag in a sibling product's codebase (38 files scanned).

## What it checks

| Check | What it catches |
|---|---|
| Invisible Unicode | Prompt-injection payloads hidden in homoglyph/zero-width chars |
| Dangerous patterns (MCP001–006) | Dynamic imports, subprocess shells, credential exfil shapes |
| Secrets | Hardcoded API keys and tokens |
| Typosquats | Package names one edit-distance from popular packages |

## Try it in 30 seconds — no signup

```bash
curl -X POST https://trust-scan-production.up.railway.app/v1/scan \
  -H "Content-Type: application/json" \
  -d '{"path": "/path/to/mcp-server-or-skill"}'
```

## Install (MCP clients)

```json
{
  "mcpServers": {
    "trust-scan": {
      "url": "https://trust-scan-production.up.railway.app/mcp/"
    }
  }
}
```

Tools: `trust_scan_server` (directory/package), `trust_scan_file` (single file).

Agent-facing API reference: [`llms.txt`](https://trust-scan-production.up.railway.app/llms.txt)

## Architecture

- Scanner (`trust_scan.py`): severity-weighted findings, Levenshtein typosquat
  detection, invisible-codepoint analysis
- REST API (`api_server.py`, FastAPI): `/health`, `/v1/scan`, `/stats`,
  hosted streamable-http MCP at `/mcp/`

## Contact

entradox@icloud.com

More