> Parel documentation. Index of every page: https://parel.si/docs/llms.txt. This page in HTML: https://parel.si/docs/tools/map_site

# map_site

> Every URL on a website, or under one path, in one call: the pages to read, before reading any of them. ₹0.20 per call; a site with no URLs is ₹0.

## What it does

The URLs of a website in one call, from the sources that already list them (its sitemaps, the links on its home page and public web indexes), in `data.rows[]` with `url`, `sources`, `last_seen` and `depth`; `data.meta` says how many pages were read (`pages_fetched`) and whether a source failed (`partial`, `warnings`). `include` keeps only URLs under a path (`"/team"`) or matching a glob (`"/blog/*"`); `max_urls` caps the list (default and maximum 5,000); `mode: "full"` adds the Wayback Machine and up to twenty seconds.

Use it before reading anything. When a list spans several pages (a team page, a blog, a catalogue), the map hands back every page's address in one call, and the next step is [`read_pages`](https://parel.si/docs/tools/read_pages.md) for up to ten of those pages in one call, or [`extract_structured`](https://parel.si/docs/tools/extract_structured.md) for the entries as a table. It lists addresses only: it does not read the pages ([`read_page`](https://parel.si/docs/tools/read_page.md) does), and it knows only the URLs the site publishes. In the live check on a 98-person team page split over five pages, the map returned the 99 person pages under `/team` and not the `/team/page/2/` to `/page/5/` list pages, which are not in the site's sitemap.

India: any public site; send the host as a bare domain.

## Inputs

Generated from the contract captured from staging on 2026-11-05.

| Field | Type | Required | Description |
|---|---|---|---|
| `mode` | `fast` or `full` | no | full adds the Wayback Machine, up to 20 s more. default "fast" |
| `domain` | string | yes | the site's host; a leading https:// is dropped, www and the bare host are one site. e.g. "example.com" |
| `include` | string | no | keep only matching URLs: a path prefix (/docs) or a glob with * over the path |
| `max_urls` | integer | no | default 5000; max 5000 |
| `include_subdomains` | boolean | no | default false |

## What drives the cost

Nothing you can turn: flat per call whatever the size of the site or `max_urls`. The source bills by the pages its scan reads (robots.txt, the home page, each sitemap, each index reply); the live call read 16 pages and the source charged under a tenth of what is shown, so a call usually costs less than the price shown; a site with no URLs is `class: "ok_free"`, ₹0. Usually five to thirty seconds; a large site can take over a minute, so the timeout is 120 s. A repeat within a week is answered from the source's cache.

## Price today

| | |
|---|---|
| Price today | **₹0.20** per call (`cost_paise: 20`) at ₹93.93 per US dollar on 17 November 2026 |
| Hold before the call | ₹0.38 (`hold_paise: 38`); when the call settles the real price is charged and the unused part is released |
| ₹0 when | it fails, or the provider answers but there is nothing in it (an empty list, no email, no page): `class: "ok_free"`, ₹0 |
| Timeout | 180 s, then `upstream_timeout` (₹0, retryable) |
| Over MCP | `run_tool` with `tool: "map_site"` |
| Over REST | `POST https://api.parel.si/v1/run` with `"tool": "map_site"` |
| Confirmation | not needed (the estimate is under the ₹200 threshold) |

## Request

```bash
curl https://api.parel.si/v1/run \
  -H "Authorization: Bearer $PAREL_KEY" \
  -H "Content-Type: application/json" \
  -d '{"tool":"map_site","input":{"domain":"americanindustrial.com","include":"/team","max_urls":200}}'
```

## What comes back

`data.rows[]` and `data.meta`. A real answer recorded during the live check of 2026-11-04 (americanindustrial.com under `/team`: 100 rows in 22 s), trimmed to three rows:

```json
{
  "meta": {
    "blocked": [],
    "elapsed_ms": 19796,
    "from_cache": false,
    "llm_calls": 0,
    "pages_fetched": 16,
    "partial": false,
    "warnings": [
      "module: site-scan: CC-MAIN-2026-43-index gave no urls (status 0 unavailable, 0 bytes)",
      "module: site-scan: CC-MAIN-2026-39-index gave no urls (status 0 unavailable, 0 bytes)",
      "module: site-scan: commoncrawl 0 · links 53 · sitemap 277 · union 100"
    ]
  },
  "recipe": "site-scan",
  "rows": [
    {
      "depth": 1,
      "last_seen": "2026-10-29T02:02:39+00:00",
      "sources": [
        "sitemap",
        "links"
      ],
      "url": "https://americanindustrial.com/team"
    },
    {
      "depth": 2,
      "last_seen": "2026-06-01T16:41:54+00:00",
      "sources": [
        "sitemap"
      ],
      "url": "https://americanindustrial.com/team/achintya-saxena"
    },
    {
      "depth": 2,
      "last_seen": "2026-06-02T21:25:08+00:00",
      "sources": [
        "sitemap"
      ],
      "url": "https://americanindustrial.com/team/adam-traina"
    },
    "… 97 more item(s) truncated"
  ],
  "version": 1
}
```

## Notes

- Flat per call. include narrows to a path prefix ("/team") or a glob ("/blog/*"); max_urls caps the list (default 5,000, the maximum here). No URLs is free.
- Only URLs the site publishes: sitemaps, home-page links and public indexes. Paginated list pages (/page/2/) are often absent while every item's own page is present.
- Usually five to thirty seconds; a large site can take over a minute, so the timeout is generous. A repeat within a week is answered from the source's cache.
- Each row: url, sources (sitemap, links, commoncrawl), last_seen, depth.

## Siblings

- [`read_pages`](https://parel.si/docs/tools/read_pages.md) to read up to ten of the URLs in one call.
- [`extract_structured`](https://parel.si/docs/tools/extract_structured.md) to turn one of the pages into a table.
- [`read_page`](https://parel.si/docs/tools/read_page.md) for one page as Markdown.
