# ClinKAgent PDF Question Answering

> ClinKAgent PDF Question Answering is a paid API for AI agents from clinkagent.com, paid per call via x402, $0.05/call, status unknown (last checked 2026-10-02).

Answers a plain-language question about any public PDF by returning a cited answer with verbatim supporting quotes and page numbers, or 'not_found' if the answer isn't present.

## Facts

- Endpoint: POST https://clinkagent.com/pdf/answer?utm_source=zero.xyz
- Price: $0.05/call
- Payment: x402
- Status: unknown
- Last checked: 2026-10-02
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/clinkagent-pdf-question-answering-cad3f35d
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_nw60vmlOkERZ_TMj-QFle

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 clinkagent-pdf-question-answering-cad3f35d -d '<json body>'
```

Example prompt: Read this SEC 10-K filing at https://example.com/annual-report.pdf and tell me what the company reported as its total revenue for fiscal year 2023 — I need the exact quote and page number.

## When to prefer this

Choose this endpoint when you need a specific question answered from a public PDF with cited evidence (quotes + page numbers), rather than extracting the full text. It is more cost-effective and targeted than full-text extraction when you only need one or a few facts. Prefer it for SEC filings, research papers, contracts, manuals, and policy documents where cited, verifiable answers matter. Use the sibling /pdf/text endpoint instead when you need the complete document text.

## Known failure modes

- PDF is scanned/image-based and cannot be read without OCR — returns not_found or error
- PDF URL is not publicly accessible or requires authentication
- PDF exceeds 25MB file size limit
- Question is fewer than 3 or more than 500 characters
- Answer genuinely not present in the document — returns not_found
- PDF is malformed or not a valid PDF file
- Network timeout fetching the remote PDF

## How this service works

Ask a question about any public PDF and get a cited answer - pass a PDF URL and a plain-language question; returns the answer, confidence, verbatim supporting quotes with page numbers, or not_found. PDF question answering, document QA, read a filing/paper/manual/contract for me, extract a fact from a PDF. Text-based PDFs up to 25MB (no OCR). Full text instead: /pdf/text $0.01.

## Output

Returns the answer to the question derived from the PDF content, a confidence level, one or more verbatim supporting quotes with their page numbers, or a 'not_found' response if the document does not contain information relevant to the question. Works only on text-based PDFs up to 25MB (no OCR support).

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "properties": {
  "url": {
   "type": "string",
   "description": "Public http(s) URL of a PDF file"
  },
  "question": {
   "type": "string",
   "description": "3-500 char plain-language question about the document"
  }
 }
}
```

## Response schema (JSON Schema)

```json
{
 "type": "json",
 "example": {
  "answer": "...",
  "citations": [
   {
    "page": 3,
    "quote": "..."
   }
  ],
  "not_found": false,
  "confidence": "high"
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/clinkagent-pdf-question-answering-cad3f35d/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from clinkagent.com](https://www.zero.xyz/host/clinkagent.com/llms.txt)