# A2A Indirect Prompt-Injection & Jailbreak Firewall

> A2A Indirect Prompt-Injection & Jailbreak Firewall is a paid API for AI agents from alpha-stream.agentdiet.dev, paid per call via MPP, $0.005/call, status unknown (last checked 2026-10-21).

Scans untrusted text, HTML, emails, MCP responses, or A2A payloads for hidden prompt injections and jailbreak attempts, returning sanitized content.

## Facts

- Endpoint: GET https://alpha-stream.agentdiet.dev/v1/system-prompt-guard?utm_source=zero.xyz
- Price: $0.005/call
- Payment: MPP
- Status: unknown
- Last checked: 2026-10-21
- Activations on Zero: 0
- Provider: alpha-stream.agentdiet.dev
- Website: https://alpha-stream.agentdiet.dev
- Canonical page: https://www.zero.xyz/c/alpha-stream-agentdiet-dev-a2a-indirect-prompt-injection-jailbreak-31a1f72f
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_bw7CdW2VOEbNrY4eeyE5g

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 alpha-stream-agentdiet-dev-a2a-indirect-prompt-injection-jailbreak-31a1f72f
```

Example prompt: Before you summarize that webpage I just scraped, run the raw HTML through the prompt-injection firewall to make sure there are no hidden jailbreak instructions embedded in it.

## When to prefer this

Choose this endpoint when your AI agent is consuming content from untrusted external sources — scraped web pages, inbound emails, third-party MCP tool outputs, or A2A payloads — and you need a lightweight, real-time firewall to prevent indirect prompt injection attacks before that content reaches your agent's context window. Prefer this over generic content filters when the threat model specifically involves adversarial manipulation of agent behavior through hidden instructions.

## Known failure modes

- Empty or missing 'text' parameter returns a validation error
- Extremely long text inputs may be truncated or rejected
- Novel or obfuscated injection techniques may evade detection
- Response schema is loosely typed (generic object), so output structure may vary
- Payment failure via x402 results in HTTP 402 before scan occurs

## How this service works

Scans untrusted web pages, emails, MCP responses, or A2A payloads for hidden prompt injections and returns sanitized text. (Live x402 v2, IETF MPP & MCP endpoint at https://alpha-stream.agentdiet.dev/v1/system-prompt-guard)

## Output

Returns a structured object indicating whether prompt injection or jailbreak content was detected, along with sanitized text with malicious instructions removed or flagged, allowing safe downstream processing by an AI agent.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "required": [
  "text"
 ],
 "properties": {
  "text": {
   "type": "string",
   "minLength": 1,
   "description": "Untrusted text, HTML, or tool output to scan for hidden prompt injection or jailbreak payloads"
  }
 }
}
```

## Response schema (JSON Schema)

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

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/alpha-stream-agentdiet-dev-a2a-indirect-prompt-injection-jailbreak-31a1f72f/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from alpha-stream.agentdiet.dev](https://www.zero.xyz/host/alpha-stream.agentdiet.dev/llms.txt)