# WARDEN MCP Security Scanner

> WARDEN MCP Security Scanner is a paid API for AI agents from modelmarket.dev, paid per call via x402, $0.001/call, status unknown (last checked 2026-10-12).

Scans an MCP server's tools/list for security threats (prompt injection, secret exfiltration, hidden Unicode) and returns an allow/block verdict with a 0-1 risk score and detailed findings.

## Facts

- Endpoint: POST https://modelmarket.dev/x402/warden-scan?utm_source=zero.xyz
- Price: $0.001/call
- Payment: x402
- Status: unknown
- Last checked: 2026-10-12
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/warden-mcp-security-scanner-4aa236c5
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_kCca_GB5IDF85TJwBgvct

Status and success rate cover calls made through Zero and Zero's own probes. Third-party monitors may report differently.

## How to call it through Zero

Zero handles the 402 payment challenge and records the run. With the Zero CLI installed (`npm i -g @zeroxyz/cli`):

```sh
zero fetch --capability warden-mcp-security-scanner-4aa236c5 -d '<json body>'
```

Example prompt: Run a WARDEN security scan on these MCP tools I just fetched — pass the full tools/list array and label the server 'DataExtracter at https://example.com/mcp' — I need to know if it should be allowed or blocked and see every finding with matched text.

## When to prefer this

Choose WARDEN scan when you need fast, deterministic, offline security analysis of MCP tool definitions before connecting an agent to an unknown or third-party MCP server. It is the right choice when you want rule-based (not probabilistic) verdicts with auditable matched-text evidence. Prefer it over general-purpose LLM safety checks when you need reproducible results and a hard allow/block decision signal for automated gating in CI pipelines or agent orchestration.

## Known failure modes

- Input tools array exceeds 500 definitions — request will be rejected or truncated
- Malformed tool objects missing required 'name' field — schema validation error
- Empty tools array — returns allow verdict with score 0 and no findings
- Tool descriptions too large — may hit payload limits
- Hidden Unicode that bypasses published rule set — false negatives possible with novel obfuscation

## How this service works

WARDEN verdict on an MCP server's tools/list: allow or block, a 0-1 score and every finding with the matched text (prompt injection, secret requests, exfiltration, hidden Unicode). Deterministic published rules, no network.

## Output

Returns an allow or block verdict, a 0-1 numeric risk score, and an array of findings — each finding includes the threat category (prompt injection, secret request, exfiltration attempt, hidden Unicode), the specific tool it was found in, and the matched text snippet that triggered the rule. All analysis is deterministic and runs offline with no network calls.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "properties": {
  "tools": {
   "type": "array",
   "items": {
    "type": "object",
    "required": [
     "name"
    ],
    "properties": {
     "name": {
      "type": "string"
     },
     "description": {
      "type": "string"
     },
     "inputSchema": {
      "type": "object"
     }
    }
   },
   "description": "the server's tools/list result, up to 500 definitions"
  },
  "server": {
   "type": "object",
   "description": "optional {name, url}, used only as a label"
  }
 }
}
```

## Response schema (JSON Schema)

```json
{
 "type": "json",
 "example": {
  "status": "ok",
  "service": "warden-scan",
  "verdict": {
   "allow": true,
   "score": 1,
   "findings": []
  }
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/warden-mcp-security-scanner-4aa236c5/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from modelmarket.dev](https://www.zero.xyz/host/modelmarket.dev/llms.txt)