# AI Agent Payment Safety Stack – Approval Unit Builder

> AI Agent Payment Safety Stack – Approval Unit Builder is a paid API for AI agents from ai-agent-payment-safety-stack.onrender.com, paid per call via x402, $0.05/call, status unknown (last checked 2026-09-15).

Builds a minimal human decision contract (approval unit) from AI-generated findings, patches, payment requests, or decision-support outputs, enabling human-in-the-loop oversight before consequential actions.

## Facts

- Endpoint: POST https://ai-agent-payment-safety-stack.onrender.com/api/approval-unit/build
- Price: $0.05/call
- Payment: x402
- Status: unknown
- Last checked: 2026-09-15
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/ai-agent-payment-safety-stack-approval-unit-builder-ded59353
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_gH-FTHhwfZdbIu2UXZR8N

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 ai-agent-payment-safety-stack-approval-unit-builder-ded59353 -d '<json body>'
```

Example prompt: Build a human approval unit for this AI-generated payment request: my agent wants to send 50 USDC to vendor 0xABC for invoice #1042 — package it as a minimal decision contract so a human can review and approve or reject it before the funds move.

## When to prefer this

Choose this endpoint when an AI agent needs to pause before executing a consequential action (especially payments or code deployment) and requires a structured, minimal human-review contract to be built from its outputs. Prefer this over ad-hoc logging or manual summaries when you need a standardized, machine-readable approval artifact that human operators or oversight dashboards can consume.

## Known failure modes

- Missing or malformed input payload returns 400 Bad Request
- Payment of 0.05 USDC not included or rejected returns 402 Payment Required
- Unsupported content type in the request body returns 415
- Service unavailable on Render cold start returns 503 with latency spike
- Oversized or complex AI output exceeds payload limits returning 413

## How this service works

Build a minimal human decision contract from AI-generated findings, patches, payment requests, or decision-support outputs.

## Output

Returns a structured approval unit (human decision contract) containing the AI-generated input (finding, patch, payment request, or recommendation) formatted as a minimal, reviewable payload that a human can inspect and approve or reject before the AI agent proceeds.

## Example request

```json
{
 "title": "Security Patch Deployment Approval",
 "summary": "AI agent has identified a critical security patch for production deployment. This approval unit packages the deployment decision for human review before execution.",
 "task_id": "task_deploy_patch_20240115",
 "agent_id": "ai_agent_security_001",
 "severity": "high",
 "source_id": "ai_agent_security_001",
 "finding_id": "finding_security_patch_20240115",
 "risk_level": "medium",
 "source_ids": [
  "patch_repository_main"
 ],
 "cost_impact": {
  "deployment_time_minutes": 15,
  "estimated_downtime_seconds": 0
 },
 "source_type": "ai_finding",
 "evidence_ids": [
  "cve_scan_report_001",
  "regression_test_suite_001"
 ],
 "test_results": [
  "unit_tests_passed",
  "integration_tests_passed",
  "security_scan_passed"
 ],
 "approver_role": "human_approver",
 "reviewer_role": "security_reviewer",
 "regression_risk": "low",
 "approval_unit_type": "payment_request",
 "rollback_available": true,
 "recommended_decision": "approve",
 "allowed_human_actions": [
  "approve",
  "reject",
  "request_rework",
  "request_more_evidence",
  "escalate"
 ],
 "suggested_human_actions": [
  "review_patch_details",
  "verify_test_results",
  "check_deployment_window"
 ],
 "blocked_actions_until_approval": [
  "deploy_to_production",
  "execute_patch"
 ]
}
```

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "properties": {
  "approval_question": {
   "type": "string"
  },
  "approval_unit_hash": {
   "type": "string"
  },
  "chain_anchor_status": {
   "type": "string"
  },
  "recommended_human_action": {
   "type": "string"
  }
 }
}
```

## Response schema (JSON Schema)

```json
{
 "type": "json",
 "example": {
  "approval_question": "Approve this security patch for staging merge?",
  "approval_unit_hash": "sha256:...",
  "chain_anchor_status": "not_anchored",
  "recommended_human_action": "approve_staging_only"
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/ai-agent-payment-safety-stack-approval-unit-builder-ded59353/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from ai-agent-payment-safety-stack.onrender.com](https://www.zero.xyz/host/ai-agent-payment-safety-stack.onrender.com/llms.txt)