# Local Document Answer or Insufficient Evidence

> Local Document Answer or Insufficient Evidence is a paid API for AI agents from market.datapackvibe.com, paid per call via x402, $0.01/call, status unknown (last checked 2026-10-15).

Answers a question using up to three supplied text passages via a local Qwen model, or explicitly returns an insufficient-evidence result if the passages don't support an answer.

## Facts

- Endpoint: POST https://market.datapackvibe.com/x402/assistant-local-evidence-answer?utm_source=zero.xyz
- Price: $0.01/call
- Payment: x402
- Status: unknown
- Last checked: 2026-10-15
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/local-document-answer-or-insufficient-evidence-5844c3d7
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_YSSrsjE7lnVDbG3Sm6ZtF

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 local-document-answer-or-insufficient-evidence-5844c3d7 -d '<json body>'
```

Example prompt: Using the local evidence-answer service, check whether this passage — 'Refunds may be requested within thirty days of purchase.' — answers the question 'How long is the refund period?' and give me the exact supporting quote or tell me there's insufficient evidence.

## When to prefer this

Choose this endpoint when you need grounded, passage-bounded question answering with explicit evidence verification — especially when you want the system to admit insufficient evidence rather than hallucinate. Prefer it over general LLM QA when auditability and exact quote matching matter, and when your documents are short excerpts (up to 3 passages of 600 chars each). Not suitable for full-document analysis, web search, or open-domain QA.

## Known failure modes

- Question exceeds 300 characters — request rejected
- Document passage exceeds 600 characters — passage rejected
- Fewer than 1 or more than 3 documents supplied — validation error
- No passage contains a quote matching the answer — returns insufficient-evidence result
- Model output may require human review as quote matching does not prove entailment
- Network or model inference failure returns a non-200 error

## How this service works

Local Document Answer or Insufficient Evidence: Use local Qwen to answer from up to three supplied passages of 600 characters. Verify exact supporting quote occurrences, or return an explicit insufficient-evidence result. Quote matching does not prove entailment or source truth; model output needs review. 0.01 USDC includes one bounded document-ai report.

## Output

Returns a structured document-AI report containing either a direct answer grounded in an exact supporting quote found within the supplied passages, or an explicit insufficient-evidence result indicating the passages do not support an answer to the question. The output is bounded to the provided text and does not hallucinate beyond it.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "properties": {
  "question": {
   "type": "string",
   "default": "How long is the refund period?",
   "maxLength": 300,
   "minLength": 1
  },
  "documents": {
   "type": "array",
   "items": {
    "type": "string",
    "maxLength": 600,
    "minLength": 1
   },
   "default": [
    "Refunds may be requested within thirty days of purchase."
   ],
   "maxItems": 3,
   "minItems": 1
  }
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/local-document-answer-or-insufficient-evidence-5844c3d7/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from market.datapackvibe.com](https://www.zero.xyz/host/market.datapackvibe.com/llms.txt)