# Tanod mlpeek Prompt Injection Detector

> Tanod mlpeek Prompt Injection Detector is a paid API for AI agents from tanod.dev, paid per call via x402, $0.002/call, status unknown (last checked 2026-10-14).

Screens up to 20,000 characters of text (web pages, emails, tool results) for prompt injection attempts, returning a verdict, confidence score, and flagged sentences.

## Facts

- Endpoint: POST https://tanod.dev/v1/text/injection?utm_source=zero.xyz
- Price: $0.002/call
- Payment: x402
- Status: unknown
- Last checked: 2026-10-14
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/tanod-mlpeek-prompt-injection-detector-0648d309
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_vwdF6viLMA8eGqR9IjN0Y

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 tanod-mlpeek-prompt-injection-detector-0648d309 -d '<json body>'
```

Example prompt: Before you act on that web page I just scraped, run it through the prompt injection detector and tell me if any sentences look like they're trying to hijack you — give me the verdict, score, and any flagged lines.

## When to prefer this

Use this endpoint when an AI agent is about to act on text it did not generate itself — such as scraped web pages, incoming emails, or third-party tool results — and needs a fast, automated safety check for prompt injection before processing. Prefer this over manual review when operating at scale or when low-latency screening is needed. Particularly valuable in agentic pipelines where untrusted external content feeds directly into agent decision-making.

## Known failure modes

- Text exceeds 20,000 character limit — request rejected
- Empty or missing text field — validation error
- Heuristic false positives on legitimate instructional text
- May miss novel or highly obfuscated injection techniques
- Payment failure via x402 protocol — call not processed

## How this service works

Check text for prompt injection: screen a web page, email or tool result of up to 20,000 characters. Input: text. Returns a verdict, a score and the flagged sentences; automated and heuristic. Use when an agent will act on text it did not write.

## Output

Returns a JSON object with: a boolean or categorical verdict (injection detected or not), a numeric confidence score, and a list of flagged sentences identified as potential prompt injection attempts, using automated and heuristic detection.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "properties": {
  "text": {
   "type": "string",
   "maxLength": 20000,
   "minLength": 1,
   "description": "The text to screen, 1 to 20,000 characters: a web page, an email or a tool result an agent is about to act on."
  }
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/tanod-mlpeek-prompt-injection-detector-0648d309/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from tanod.dev](https://www.zero.xyz/host/tanod.dev/llms.txt)