Niches · Web search APIs

Read a known URL as Markdown

One company's docs show a full fit.

1companies, full fit
12companies, fit with a limit
31companies carry the tag
12share of the category's work
7first-hand accounts

What counts as done

Main content of the page as Markdown with links kept, with a distinct error (not an empty body) when the page is blocked or needs JavaScript.

Measured, for our own tests: under 10 seconds. No docs we read state these figures, so they don't count against a listing here.

How agents ask for it: read this page, fetch this URL and summarise it, what does this link say

Does the job 1

The docs show it meets the test above, and nothing we read says otherwise.

ListingWhy
Parallel Search and Task APIs BB
Parallel Web Systems Inc.
/v1/extract returns markdown with links for JavaScript-heavy pages and PDFs; fetch_policy sets max_age_seconds and disable_cache_fallback makes a failed live fetch an error. source

Does the job, with a limit 12

It does the job, but something in the docs or a user's account will get in an agent's way. The limit is named in each row.

ListingWhy
DeepSeek Search and Fetch API C
Hangzhou DeepSeek Artificial Intelligence Co., Ltd.
Fetch renders JavaScript with a Firefox user agent from Huawei Cloud Hong Kong, but one tester got text only with hyperlinks dropped. source
Exa API + MCP B
Exa Labs Inc.
Contents takes URLs with text, maxCharacters, maxAgeHours 0 for a live fetch and livecrawlTimeout, but the pages read do not say whether JavaScript is rendered or links are kept. source
Fetch (MCP reference server) B
Model Context Protocol, a Series of LF Projects, LLC
HTML to Markdown via Readability with max_length and start_index paging and isError on robots refusals and HTTP errors, but no JavaScript rendering (progressively rendered content dropped) and Readability was skipped on Windows until October 2026. source
Kagi Search API C
Kagi Inc.
Page Markdown comes through the extract option; the MCP kagi_extract tool returned no content for one user in October 2026 (closed 28 October), and rendering is not described. source
Linkup B
Linkup Technologies SAS
/fetch returns markdown with renderJs and a pro mode, and a platform runs 100-plus fetches a job on it, but the error behaviour for blocked pages was not in the pages read. source
Perplexity Search API BB
Perplexity AI, Inc.
fetch_url exists only as an Agent API tool at $0.50 per 1,000 plus model tokens, with failure markers for robots blocks and no_result_returned; no standalone endpoint. source
SERPHouse E
Patoliya Infotech LLP
Scraper API /scrape/md returns Markdown from the rendered page at 1 to 10 credits by site type, but no error behaviour for blocked pages is documented. source
SerpKite BB
Anastasia Klimova, trading as SerpKite
/v1/webpage returns clean Markdown and metadata for HTML or PDF, and /v1/extract lists failed URLs at no charge, but JavaScript rendering is not stated. source
Tavily API + MCP A
AlphaAI Technologies Inc. dba Tavily
/extract returns markdown with per-URL failed_results and an extract_depth, but Mintlify docs sites came back as empty results with no error and no maintainer reply. source
TinyFish Search B
Tiny Fish, Inc.
Fetch returns Markdown by default with per-URL errors and timeouts, but links are kept only as a separate list when links=true, and rendering is implied rather than stated. source
Valyu BB
Valyu.Network Ltd
A Contents API extracts text from up to 50 URLs at 0.2 credits each, paying only for successes, but its guide was not opened so rendering and errors are unread. source
You.com APIs BB
SuSea, Inc., d/b/a you.com
Contents API returns Markdown or HTML for up to 10 URLs with crawl_timeout (20 to 30 seconds suggested for JavaScript-heavy pages), but per-URL error behaviour was not in the overview read. source

Couldn't check 1

The docs were closed to us, or silent. These carry no mark either way.

ListingWhy
SpecStream B
SpecStream SAS
POST /v1/extract is listed at $0.80 per 1,000 pages, but its docs page was not opened.

Doesn't do this job 12

Listings in the category that do something else, with the reason.

ListingWhy
Web Search Tool on Amazon Bedrock AgentCore BB
Amazon Web Services, Inc. (regional AWS entities by account location)
Web Search Tool takes a query only; no URL fetch tool is part of this connector.
Arcmira B
Arcmira, Inc.
arcmira_youtube_transcript reads a video transcript from its link; no general URL fetch.
Grounding with Bing Search B
Microsoft Corporation
No URL fetch tool in the Bing grounding tools.
Brave Search API + MCP B
Brave Software, Inc.
No fetch endpoint; LLM Context starts from a query, not a URL.
Cloudflare Web Search API B
Cloudflare, Inc.
Search only; no fetch or extract route in the Web Search API.
Grep AI B
Parcha Labs, Inc.
No fetch endpoint.
Liner Search MCP E
Liner
No fetch endpoint.
Mojeek Web Search API F
Mojeek Limited
No fetch endpoint.
NewsCatcher B
NewsCatcher, Inc.
No fetch endpoint.
SearchAPI.io D
SearchApi, LLC
No fetch endpoint.
SerpApi B
SerpApi, LLC
No fetch endpoint.
Serper F
Serper
No fetch endpoint in the listing's capabilities or on the home page.

What stops agents finishing

  • empty body on JavaScript sites with no fallback to a browser
  • robots or bot check refusal with no solver
  • per-domain limits that are not published (DeepSeek Fetch)

Around it

Before: search.ground-fact. After: document-extraction.table-from-text.

Where this comes from

Weight 12 of 100 in Web search APIs: 7 of 44 accounts: DeepSeek fetch reviews, Kagi extract no-content issue, Tavily extract empty on SPA, Linkup fetch pricing, Firecrawl fallback, Fetch reference server issues

The same as data: /api/v1/use-cases/search.fetch-url.json, or the MCP tool get_use_case with id search.fetch-url. Other jobs in this category: /api/v1/niches?category=search.

Draft, read 2026-11-12 from public docs and first-hand accounts, not yet reviewed by a person. A mark says what the docs show, not what a run measured.

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.