# Penny Engine — AI Agent Outcome Guard

> Penny Engine — AI Agent Outcome Guard is a paid API for AI agents from penny-paid-batch-production.up.railway.app, paid per call via x402, $0.001/call, status unknown (last checked 2026-10-23).

Validates AI agent output against declared expectations, returning a deterministic PASS/FAIL verdict and SHA-256 fingerprint of the result.

## Facts

- Endpoint: POST https://penny-paid-batch-production.up.railway.app/api/proof-gate?utm_source=zero.xyz
- Price: $0.001/call
- Payment: x402
- Status: unknown
- Last checked: 2026-10-23
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/penny-engine-ai-agent-outcome-guard-ffac1b70
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_XRJcqO7w0W2K79TjBqkXT

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 penny-engine-ai-agent-outcome-guard-ffac1b70 -d '<json body>'
```

Example prompt: Check if this agent result passes my expectations — it should have a non-empty 'summary' field, a 'confidence' value between 0.8 and 1.0, and the 'status' field should equal 'completed'. Give me a PASS or FAIL and the SHA-256 fingerprint.

## When to prefer this

Choose this endpoint when you need a lightweight, pay-per-call, cryptographically auditable quality gate for structured AI agent outputs — especially in automated M2M pipelines where you need a tamper-evident fingerprint alongside a binary verdict. It is ideal when you want deterministic validation rules (field existence, value equality, numeric ranges) rather than semantic or LLM-based evaluation. At $0.001 USDC per call it is cost-effective for high-frequency agent monitoring.

## Known failure modes

- Missing or malformed 'result' object causes validation to fail with an error response
- Empty or invalid 'expectations' object may result in zero checks and an ambiguous outcome
- Payment failure via x402 returns HTTP 402 and no validation is performed
- Malformed JSON in the request body returns an HTTP 400 error
- Network timeout or Railway infrastructure unavailability returns HTTP 5xx

## How this service works

Machine-to-machine result validation via x402. One structured call, deterministic PASS or FAIL and SHA-256 fingerprint, $0.001 USDC on Base.

## Output

Returns a JSON object with a boolean 'passed' field, a 'verdict' string ('PASS' or 'FAIL'), an 'evidence' object containing the hashing algorithm name ('SHA-256') and the SHA-256 hash of the submitted result, and a 'checksTotal' integer indicating how many validation checks were evaluated.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "properties": {
  "result": {
   "type": "object"
  },
  "expectations": {
   "type": "object",
   "properties": {
    "equals": {
     "type": "object"
    },
    "numberRanges": {
     "type": "object"
    },
    "nonEmptyPaths": {
     "type": "array",
     "items": {
      "type": "string"
     }
    },
    "requiredPaths": {
     "type": "array",
     "items": {
      "type": "string"
     }
    }
   }
  }
 }
}
```

## Response schema (JSON Schema)

```json
{
 "type": "json",
 "example": {
  "passed": true,
  "service": "M2M Agent Outcome Guard",
  "verdict": "PASS",
  "evidence": {
   "algorithm": "SHA-256",
   "resultSha256": "example_only"
  },
  "checksTotal": 4
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/penny-engine-ai-agent-outcome-guard-ffac1b70/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from penny-paid-batch-production.up.railway.app](https://www.zero.xyz/host/penny-paid-batch-production.up.railway.app/llms.txt)