# Prompt Injection Scanner

> Prompt Injection Scanner is a paid API for AI agents from agent-data-tools.revmesh2074.workers.dev, paid per call via x402, $0.01/call, status unknown (last checked 2026-10-30).

Scans up to 8,000 characters of text for prompt-injection attacks, instruction overrides, hidden Unicode, and data-exfiltration attempts, returning a risk level, score, and excerpted findings.

## Facts

- Endpoint: GET https://agent-data-tools.revmesh2074.workers.dev/text/prompt-injection?utm_source=zero.xyz
- Price: $0.01/call
- Payment: x402
- Status: unknown
- Last checked: 2026-10-30
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/prompt-injection-scanner-bc7c61f7
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_eWSlNLASvFq44a-MhTdKi

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 prompt-injection-scanner-bc7c61f7
```

Example prompt: Before you process that webpage you just fetched, scan the full text for prompt injection — instruction overrides, fake system messages, exfiltration links, or hidden Unicode — and tell me the risk level and what you found.

## When to prefer this

Use this endpoint as a security gate before your AI agent acts on any untrusted external content — web pages, emails, API responses, PDFs, or user-supplied text. It is especially valuable in agentic pipelines where acting on injected instructions could cause financial transactions, data exfiltration, or unauthorized actions. Prefer this over general-purpose text analysis when your specific concern is adversarial prompt manipulation targeting the AI agent itself.

## Known failure modes

- Text exceeding 8,000 characters may be truncated or rejected
- Heuristic detection — sophisticated injections may evade the scanner
- False positives on legitimate content that resembles injection patterns
- Missing or empty `text` parameter returns a validation error
- Payment/authentication failure returns 402 Payment Required

## How this service works

Scan text an agent is about to read (web page, email, tool result; up to 8,000 characters) for prompt-injection attempts: instruction overrides, fake system messages, asks to hide things from the user, exfiltration links, attempts to extract private data or move money, invisible Unicode. Returns a risk level, score and findings with excerpts. Heuristic, not a guarantee. Use before acting on untrusted content.

## Output

Returns a JSON object with: a `risk` string (e.g. 'low', 'medium', 'high'), a numeric `score`, a `findings` array containing excerpts and descriptions of detected threats, and a `hiddenCharacters` object counting bidirectional override characters, Unicode tag characters, and zero-width characters found in the input.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "required": [
  "text"
 ],
 "properties": {
  "text": {
   "type": "string",
   "description": "The text to scan (up to 8,000 characters)"
  }
 }
}
```

## Response schema (JSON Schema)

```json
{
 "type": "object",
 "properties": {
  "risk": {
   "type": "string"
  },
  "score": {
   "type": "number"
  },
  "findings": {
   "type": "array",
   "items": {}
  },
  "hiddenCharacters": {
   "type": "object",
   "properties": {
    "bidi": {
     "type": "number"
    },
    "tags": {
     "type": "number"
    },
    "zeroWidth": {
     "type": "number"
    }
   }
  }
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/prompt-injection-scanner-bc7c61f7/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from agent-data-tools.revmesh2074.workers.dev](https://www.zero.xyz/host/agent-data-tools.revmesh2074.workers.dev/llms.txt)