> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.beatapi.io/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.beatapi.io/_mcp/server.

# Can the AI fill in the remainder of a truncated article?

> Never extrapolate past truncated: true in POST /v1/web/read; cite only returned passages and run a follow-up Read request with a higher max_chars (up to 100,000) or a focused query.

No. Never let an AI guess or fill in text omitted after `truncated: true` in `POST /v1/web/read`. Answer only what the returned `results[].content` explicitly supports, flag missing sections as open, and send a follow-up `POST /v1/web/read` request to retrieve the remaining text.

## Fetch missing passages cleanly

Two parameters control how much text returns for each URL:

* **`max_chars`**: `500–100,000` characters per URL (default `20,000`). Raise `max_chars` when a long specification or changelog is cut off early.
* **`query`**: Optional string up to 400 characters that extracts only passages relevant to a targeted question (for example `"legacy API shutdown conditions"` when the migration date was already captured).

```bash
curl -sS https://api.beatapi.io/v1/web/read \
  -H "Authorization: Bearer $BEATAPI_API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{"urls":["https://example.com/migration"],"query":"legacy API shutdown conditions","max_chars":12000}'
```

## Interpret `truncated: false` with `query`

When `query` is set, `truncated: false` confirms only that the extracted relevant passages fit inside `max_chars`—not that the entire webpage was returned. Each follow-up Read is billed as a new call (`$0.002` per successfully read URL).

## Contracts and related questions

Before running a request, inspect the contract for free: `data:web.search`, `data:web.read`, `data:web.map`.

[Back to Web Search questions](/faq/web).