# Verdict L402 – Ask Human Decision Gate

> Verdict L402 – Ask Human Decision Gate is a paid API for AI agents from verdict.kechlab.com, paid per call via x402, $0.0092/call, status unknown (last checked 2026-10-21).

Evaluates whether an AI agent should proceed autonomously, proceed with a check, or stop and ask a human before taking an action, based on irreversibility, ambiguity, missing info, and declared stakes.

## Facts

- Endpoint: POST https://verdict.kechlab.com/v1/ask-human?utm_source=zero.xyz
- Price: $0.0092/call
- Payment: x402
- Status: unknown
- Last checked: 2026-10-21
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/verdict-l402-ask-human-decision-gate-dda2ff22
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_2BEwwRGoOnZfTLAlH3jSg

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 verdict-l402-ask-human-decision-gate-dda2ff22 -d '<json body>'
```

Example prompt: Before I go ahead and delete all the flagged records from the database, can you check whether this is safe to do autonomously or if I should pause and ask the user first? The task is 'clean up stale entries', stakes are high, and I know the records haven't been backed up yet.

## When to prefer this

Use this endpoint when an autonomous agent is about to take an action that could be irreversible, ambiguous, or high-stakes, and you want a programmatic signal on whether to proceed or escalate to a human. Prefer this over hardcoded rules when the stakes or ambiguity level varies per task. It is specifically designed for agentic workflows that need a lightweight human-in-the-loop gate without building custom decision logic.

## Known failure modes

- Missing required 'plan' or 'task' fields returns a 400 validation error
- Overly vague plan or task text may produce a conservative 'ask_human' verdict even for safe actions
- Incorrect stakes enum value (not 'low', 'normal', or 'high') causes a schema rejection
- Network timeout or service unavailability blocks the gate check, requiring fallback logic
- Very short or context-free inputs may yield low-confidence verdicts with minimal reasons

## How this service works

For autonomous agents: before acting, continue or stop and ask? Estimates irreversibility, ambiguity of the instruction, missing information and risk, scaled to the stakes you declare. Returns proceed, proceed_with_check or ask_human with reasons.

## Output

Returns one of three verdicts: 'proceed' (safe to act autonomously), 'proceed_with_check' (act but flag for review), or 'ask_human' (stop and seek clarification or approval), accompanied by structured reasons explaining the irreversibility, ambiguity, missing information, and risk factors that drove the decision.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "properties": {
  "plan": {
   "type": "string",
   "description": "what the agent is about to do"
  },
  "task": {
   "type": "string",
   "description": "the instruction received by the agent"
  },
  "stakes": {
   "enum": [
    "low",
    "normal",
    "high"
   ],
   "type": "string",
   "description": "(optional) low, normal or high: stakes of a mistake (normal by default)"
  },
  "context": {
   "type": "string",
   "description": "(optional) what the agent knows"
  }
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/verdict-l402-ask-human-decision-gate-dda2ff22/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from verdict.kechlab.com](https://www.zero.xyz/host/verdict.kechlab.com/llms.txt)