# guard402

> Reliability layer for AI agents. Content Firewall, Output Validator, and Loop Detector. Pay-per-call with USDC via x402.

## Overview

guard402 provides three API endpoints that catch the problems agents can't catch themselves:

- Agents trust malicious input → **Content Firewall** scans for prompt injection
- Agents hallucinate facts → **Output Validator** cross-checks against context
- Agents get stuck in loops → **Loop Detector** identifies stuck patterns

No API keys. No accounts. Pay per call with USDC on Base or Solana via the x402 protocol.

## API Base URL

https://api.guard402.com

## Endpoints

### POST /scan — Content Firewall ($0.001)
Scans content for prompt injection attacks. Regex fast path + LLM fallback.
Input: { "content": "...", "content_type": "text|html|markdown" }
Output: { "safe": bool, "score": 0-1, "threats": [...], "sanitized": "..." }

### POST /validate — Output Validator ($0.005)
Validates agent output against task and context for hallucinations.
Input: { "task": "...", "output": "...", "context": "..." }
Output: { "valid": bool, "confidence": 0-1, "issues": [...] }

### POST /check — Loop Detector ($0.001)
Detects stuck agent patterns: repeater, looper, wanderer. Pure algorithmic, sub-ms.
Input: { "session_id": "...", "step": N, "action": "...", "params": {}, "result": "...", "goal": "..." }
Output: { "stuck": bool, "pattern": "repeater|looper|wanderer", "suggestion": "..." }

### GET /health — Free
Returns service status, uptime, version.

### GET /.well-known/agent-card.json — Free
A2A agent card for discovery.

## Payment (x402 Protocol)

1. Call any paid endpoint → receive HTTP 402 with payment details
2. Send USDC to the `payTo` address on Base or Solana
3. Retry with `X-PAYMENT` proof header → receive 200 with result

Client libraries: x402-js, x402-python, x402-rs

## Links

- Website: https://guard402.com
- Blog: https://guard402.com/blog
- Blog RSS: https://guard402.com/blog/feed.xml
- API Docs: https://guard402.com/docs
- OpenAPI Spec: https://guard402.com/openapi.yaml
- Agent Card: https://api.guard402.com/.well-known/agent-card.json
- Health: https://api.guard402.com/health