# AgentProbe v1.0.0 – Bilingual LLM-Agent Adversarial Test Pack (Case Retrieval)

> AgentProbe v1.0.0 – Bilingual LLM-Agent Adversarial Test Pack (Case Retrieval) is a paid API for AI agents from agentprobe.pythonanywhere.com, paid per call via x402, $0.003/call, status unknown (last checked 2026-09-14).

Returns a filtered set of hand-authored, OWASP-mapped adversarial test cases for LLM agents in English or Chinese, parameterised by count, category, severity, seed, and language.

## Facts

- Endpoint: GET https://agentprobe.pythonanywhere.com/v1/cases
- Price: $0.003/call
- Payment: x402
- Status: unknown
- Last checked: 2026-09-14
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/agentprobe-v1-0-0-bilingual-llm-agent-adversarial-test-pack-case-9a2f4aa2
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_TZXIc9lR-lRgWPIFiHWsM

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 agentprobe-v1-0-0-bilingual-llm-agent-adversarial-test-pack-case-9a2f4aa2
```

Example prompt: Pull 10 high-severity adversarial test cases in English from AgentProbe — specifically ones in the prompt-injection category — using seed 'run42' so the sample is reproducible.

## When to prefer this

Choose this endpoint when you need structured, OWASP-mapped adversarial prompts specifically designed to stress-test OpenAI-compatible LLM agents, especially when bilingual (EN/ZH) coverage or reproducible seed-based sampling is required. Prefer it over generic prompt datasets when you need curated, hand-authored cases with severity and category metadata for systematic red-teaming or compliance checks.

## Known failure modes

- Invalid 'lang' value (not 'en' or 'zh') returns a validation error
- 'n' exceeding the total available case count (53) may return fewer results than requested
- Unknown or misspelled 'category' or 'severity' values may return an empty result set
- Payment failure (x402) blocks access to cases
- Server unavailability on PythonAnywhere free tier may yield timeouts

## How this service works

53 hand-authored, OWASP-mapped adversarial cases (EN+ZH) plus a zero-dependency MIT harness that scores any OpenAI-compatible LLM agent. US$6.99.

## Output

A JSON object containing a filtered list of adversarial test cases. Each case includes a hand-authored adversarial prompt in the requested language (English or Chinese), its OWASP category mapping, severity level, and associated metadata. The total number of returned cases is bounded by the 'n' parameter and is reproducible when a seed is provided.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "properties": {
  "n": {
   "type": "integer"
  },
  "lang": {
   "enum": [
    "zh",
    "en"
   ],
   "type": "string"
  },
  "seed": {
   "type": "string"
  },
  "category": {
   "type": "string"
  },
  "severity": {
   "type": "string"
  }
 }
}
```

## Response schema (JSON Schema)

```json
{
 "type": "object"
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/agentprobe-v1-0-0-bilingual-llm-agent-adversarial-test-pack-case-9a2f4aa2/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from agentprobe.pythonanywhere.com](https://www.zero.xyz/host/agentprobe.pythonanywhere.com/llms.txt)