# Batcave Evidence Judge – Agent Action Reviewer

> Batcave Evidence Judge – Agent Action Reviewer is a paid API for AI agents from control-replacing-spyware-starting.trycloudflare.com, paid per call via x402, $0.02/call, status unknown (last checked 2026-09-15).

Performs independent, evidence-bound review of a proposed agent action and returns a structured pass/caution/reject verdict with reasoning.

## Facts

- Endpoint: POST https://control-replacing-spyware-starting.trycloudflare.com/v1/review-decision
- Price: $0.02/call
- Payment: x402
- Status: unknown
- Last checked: 2026-09-15
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/batcave-evidence-judge-agent-action-reviewer-5766ec79
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_nCfnlGO86g2fwHtBxoffg

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 batcave-evidence-judge-agent-action-reviewer-5766ec79 -d '<json body>'
```

Example prompt: Before you execute this action, run it through the Batcave Evidence Judge: the proposed action is 'transfer $500 to vendor account X-992', here's the evidence array with the three approval emails and invoice, the reasoning summary is 'vendor invoice matches approved PO and three sign-offs received', and the risk level is high — give me a pass, caution, or reject verdict.

## When to prefer this

Choose this endpoint when an AI agent needs an independent, evidence-bound safety gate before executing a consequential or irreversible action — especially at medium or high risk levels. It is specifically designed for multi-agent pipelines where one agent proposes an action and a separate judge validates it against supplied evidence, making it preferable to general-purpose LLM calls for this use case because it enforces structured pass/caution/reject output and evidence anchoring.

## Known failure modes

- Missing required proposed_action field returns a validation error
- Evidence array exceeding 16 items or individual items over 1000 chars will be rejected
- Invalid risk enum value (not low/medium/high) causes a schema error
- Reasoning summary over 4000 characters will be truncated or rejected
- Payment failure via x402 protocol results in 402 response and no verdict
- Empty or vague evidence may produce a caution rather than a definitive pass/reject

## How this service works

Batcave Evidence Judge: independent evidence-bound review of a proposed agent action with pass/caution/reject structured output.

## Output

A structured JSON response containing a verdict of pass, caution, or reject, along with supporting reasoning tied to the submitted evidence and proposed action. The output is designed for downstream agent logic to gate execution based on the judgment.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "properties": {
  "risk": {
   "enum": [
    "low",
    "medium",
    "high"
   ],
   "type": "string"
  },
  "task": {
   "enum": [
    "review_decision"
   ],
   "type": "string"
  },
  "job_id": {
   "type": "string",
   "maxLength": 128
  },
  "evidence": {
   "type": "array",
   "items": {
    "type": "string",
    "maxLength": 1000
   },
   "maxItems": 16
  },
  "proposed_action": {
   "type": "string",
   "maxLength": 2000,
   "minLength": 1
  },
  "reasoning_summary": {
   "type": "string",
   "maxLength": 4000
  }
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/batcave-evidence-judge-agent-action-reviewer-5766ec79/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from control-replacing-spyware-starting.trycloudflare.com](https://www.zero.xyz/host/control-replacing-spyware-starting.trycloudflare.com/llms.txt)