Search and Fetch API

Search the web and read web pages through /v1, charged to your Cloud project per request.

Search the web and read the text of web pages with your API key at https://api.dexto.ai. Each request is Dexto Cloud usage of your key’s Cloud project: it is charged from your organization’s credits, counts toward the project’s and key’s monthly spend limits, and returns what it cost in usage.charge_usd.

Use a developer API key, or the key Dexto gives a deployed Function or Site. A CLI sign-in key gets 403.

Search the web

POST /v1/search

curl -X POST https://api.dexto.ai/v1/search \
  -H "Authorization: Bearer $DEXTO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"query":"Cloudflare Workers limits","num_results":5}'

Field	Type	Default	Notes
`query`	`string`		1-2,000 characters.
`num_results`	`integer`	`8`	1-20.
`type`	`string`	`auto`	`auto`, or `fast` for lightweight lookups.
`max_characters`	`integer`	`2000`	Highlight characters per result, 100-10,000.
`max_age_hours`	`integer`		Use cached pages only when younger; `0` always reads live.

{
    "id": "8f1c6b55-7d6c-4ad6-9a8e-0f1a33a1c0de",
    "object": "search",
    "results": [
        {
            "title": "Limits · Cloudflare Workers docs",
            "url": "https://developers.cloudflare.com/workers/platform/limits/",
            "published_date": "2026-09-30",
            "highlights": ["Workers on the Paid plan can run for up to 5 minutes of CPU time..."]
        }
    ],
    "usage": { "charge_usd": 0.01625 }
}

Fetch web pages

POST /v1/fetch

curl -X POST https://api.dexto.ai/v1/fetch \
  -H "Authorization: Bearer $DEXTO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"urls":["https://developers.cloudflare.com/workers/platform/limits/"]}'

Field	Type	Default	Notes
`urls`	`string[]`		1-10 absolute `http` or `https` URLs without credentials.
`max_characters`	`integer`	`20000`	Text characters per page, 100-100,000.
`max_age_hours`	`integer`		As for search.

Each distinct URL gets one result. A page that could not be read has status: "error" and an error with a code and the page’s http_status when there was one; it is not charged.

{
    "id": "0d1c6b55-7d6c-4ad6-9a8e-0f1a33a1c0de",
    "object": "fetch",
    "results": [
        {
            "url": "https://developers.cloudflare.com/workers/platform/limits/",
            "status": "ok",
            "title": "Limits · Cloudflare Workers docs",
            "text": "Limits ...",
            "error": null
        }
    ],
    "usage": { "charge_usd": 0.00125 }
}

Pricing and errors

Search and fetch are charged at the provider’s cost plus 25%. A search with 8 results is about $0.019; a fetched page is about $0.00125. Each request is checked against your credits and spend limits before anything is fetched.

Status	Code	When
`400`	`validation_failed`	The body is invalid; nothing is charged.
`402`	`insufficient_credits`	The organization’s credits do not cover the request.
`402`	`spend_limit_reached`	The project or key reached its monthly spend limit.
`403`	`forbidden`	The key is not a developer key or a deployed key.
`429`	`rate_limited`	More than 120 requests a minute for the organization.
`502`	`upstream_failure`	The search provider failed; nothing is charged.

Cloud SDK

import { createCapabilitiesClient } from "@dexto-cloudflare/capabilities-client";

const cloud = createCapabilitiesClient({ apiKey: process.env.DEXTO_API_KEY });
const { results } = await cloud.web.search({
    query: "Cloudflare Workers limits",
});
const pages = await cloud.web.fetch({ urls: [results[0].url] });