Head to head · Fetching pages · October 2026 research run

Fetch (MCP reference server) vs SERPHouse

Fetch (MCP reference server) scores 60.1 (C) on agent readiness against SERPHouse's 42.4 (E), and leads in every scored category. Both do fetching pages.

Best web search APIs for AI agents · All 308 search comparisons

Which one, for what

Fetch (MCP reference server) C

Best for Reading a known URL as Markdown in a local or sandboxed agent, at no cost.

Ahead on

  • Reliability, 55 against 40
  • Agent ergonomics, 93 against 64
  • Security & auth, 47 against 11
  • Payments & pricing, 60 against 40
  • Maintenance & community, 72 against 23
  • Transparency & trust, 75 against 50

Also in its favour

  • No key needed to call it
  • Open source

Watch for

The address guard doesn't cover DNS rebinding or hosts only a proxy can resolve, per its README

SERPHouse E

Best for Low-cost Google, Bing and Yahoo result parsing for SEO and rank tracking, with many Google verticals.

Also in its favour

  • A hosted endpoint, with nothing to install
  • Free to start without a card
  • No incidents deducted, where Fetch (MCP reference server) loses 6 points for them

Watch for

The hosted MCP server takes the API key only in its URL path, and the REST API also accepts api_token in the query string

Why the edge

Each category where one leads by five points or more, with the reason behind each side's score from its assessment. The numbers are in the table below; this is what they stand for.

Maintenance & community Fetch (MCP reference server) +49

Fetch (MCP reference server) 72

2026.10.28, stamped and tagged v2.2.0 on 28 October 2026, 7 days before this check (30). Releases of this package in the 90 days from 6 August are 2026.8.18 (an SDK pin and a README note), 2026.10.10 (tag v1.0.0), 2026.10.19 (tag v2.0.0), 2026.10.24 (tag v2.1.0) and 2026.10.28, the last four with code changes. Dates come from the repository's tags and CalVer stamps, since pypi.org didn't resolve from our shell today. The package has no separate CLI or plugin (20). Fixes for fetch merged through October cite issues filed for them; reply times on issues couldn't be read (12). Not in the official MCP registry on 4 November (a search for server-fetch returns nothing), though the README carries an mcp-name for io.github.modelcontextprotocol/server-fetch (0). Current SDK (mcp>=2.3,<3), CI on push, a dependency sweep and a pip-audit step before each release (10). 30+20+12+0+10 = 72.

SERPHouse 23

Read for a closed service, with its repositories read as code. No changelog or dated API change was found. The newest release is MCP server 1.1.0 on 6 July 2026, 123 days before 6 November, and the last commit in its repository was on 22 July (10). No release or dated change in the 90 days from 8 August; the about page's 2026 entry (Forums, Place Details and Scraper APIs) is undated (0). No changelog; support by email. Issues weren't read, since the GitHub API refused us for its rate limit (5 of 25). Not in the official MCP registry, though the repository's server.json names io.github.SERPHouse/serphouse-mcp. Node.js SDK last changed 24 June 2026, PHP SDK 10 May 2024, and no Python SDK repository was found under the SERPHouse organisation (5 of 15). The MCP repository has an npm publish workflow and no tests (3 of 10). 10 + 0 + 5 + 5 + 3 = 23.

Security & auth Fetch (MCP reference server) +36

Fetch (MCP reference server) 47

No credentials to leak and nothing to scope, so the middle band (20). GET only. Since 2026.10.10 the server refuses private, loopback, link-local and cloud metadata addresses by default and checks the robots.txt request and every redirect hop the same way, with --allow-private-ips to turn the guard off. The README says it doesn't cover DNS rebinding between the check and the connection, or a host only a proxy can resolve (15, judgement call). It returns untrusted page text unmarked, and the README has no prompt-injection guidance (0). No call log (0). SECURITY.md, rewritten on 29 September 2026 and on main since the v1.0.0 tag, names fetch in scope, takes private reports through GitHub and publishes advisories once a fixed release is out. No bounty or certification. Graded as the filesystem and sequential-thinking servers (12). OSV holds no advisory for mcp-server-fetch. Three fixes merged on 24 October from branches named for GHSA-66qx-2266-pmj6, GHSA-3jqg-r5pp-rmp2 and GHSA-v38p-wf49-hx2j shipped in 2026.10.24. None is in OSV or in GitHub's own advisory database on 4 November, so none is published as far as can be read and none is counted (as the filesystem server's four). 20+15+0+0+12 = 47.

SERPHouse 11

One key per account, one account per user, no scopes or rotation found (10). The REST API documents api_token in the query string and the hosted MCP server takes the key only in its URL path, less 10 (0 of 30). Searches only read, with no read-only or restricted key (8 of 20). Results are third-party snippets and AI Overview text, with no prompt-injection guidance (0). An account tool shows plan and credit usage; no per-call log was found (3 of 15). No security.txt (404), disclosure policy, bug bounty or certification found (0 of 20). 0 + 8 + 0 + 3 + 0 = 11.

Agent ergonomics Fetch (MCP reference server) +29

Fetch (MCP reference server) 93

One tool, about 1,160 characters of definition, roughly 300 tokens (25). max_length and start_index page through long documents, and a truncated response names the next start_index (20). Invalid input returns INVALID_PARAMS, and a robots.txt or private-address refusal explains why and what the user can do (16). Since 2026.10.19 the tool carries readOnlyHint true, destructiveHint false, idempotentHint true and openWorldHint true, and a GET is safe to repeat (20). Only url is required and the defaults are sensible. Python only, plus a Docker image (12). 25+20+16+20+12 = 93.

SERPHouse 64

23 MCP tools with no toolsets or filtering (15 of 25). num_result, max_pages, paging, device, location, language, date and video filters (16 of 20). Status codes are tabulated, but a 200 can also mean an error and the body carries only 'Please try again' (10 of 20). Searches only read and failed requests aren't charged, but the MCP tools carry no readOnlyHint or other annotations (10 of 20). serphouse_search needs only q and defaults to the US; most other tools need loc or loc_id. Official Node.js, PHP and Python SDKs are documented (13 of 15). 15 + 16 + 10 + 10 + 13 = 64.

Transparency & trust Fetch (MCP reference server) +25

Fetch (MCP reference server) 75

The package is MIT, inside a repository moving new contributions to Apache-2.0, both OSI licences (28). Local software that keeps nothing and sends nothing beyond the request. There's no written statement, but the source is short and shows it (20). RELEASING.md describes the CalVer scheme and the README states the SDK requirement, but there's no deprecation policy or changelog for this package (8). No telemetry in the code (20). 28+20+8+20 = 76.

SERPHouse 50

Closed service under terms naming Patoliya Infotech LLP in India, and the MCP server is MIT (18 of 30). The terms give retention periods (JSON results up to 6 months, HTML up to 1 month, search history indefinitely) and say personal data sits on DigitalOcean in India, and they advise EEA users not to use the service. The privacy policy is generic and neither document carries a date (15 of 30). No deprecation policy, and the terms allow products to be discontinued without notice (0 of 20). DigitalOcean in India named for account data; no sub-processor list (6 of 20). 18 + 15 + 0 + 6 = 39.

Payments & pricing Fetch (MCP reference server) +20

Fetch (MCP reference server) 60

Free, self-hosted, nothing to buy, so 20 + 20 + 20. No payment protocol (0).

SERPHouse 40

No x402, MPP or L402 (0). Per-1,000 prices for SERP, autocomplete and top-100 requests published without login (20). Free trial of 4,000 credits for 7 days, 'No Credit Card Required' (20). A person signs up in a browser and confirms an email (0). 0 + 20 + 20 + 0 = 40.

Reliability Fetch (MCP reference server) +15

Fetch (MCP reference server) 55

Scored as a local stdio package. Official PyPI package mcp-server-fetch and the mcp/fetch Docker image, Python 3.10 or later stated (20). The monorepo's python.yml runs ruff, pyright and pytest for fetch on push, with 13 test files for this server. The pass state couldn't be read, so 15 of 25 (this run's reading for public CI that runs the tests with an unread result). The October releases close filed issues, among them #4838 (private addresses), #767, #4988, #5134, #5158, #5181, #5197 and #5199; the open issue list couldn't be read (15). CalVer versions (2026.10.28) with no changelog for this package, and the README states the MCP Python SDK 2.x requirement (5). The classifier still says 4 - Beta and the repository calls its servers reference implementations, not production-ready (0). 20+15+15+5+0 = 55.

SERPHouse 40

Read with the hosted lines. No status page is linked from the site or docs, and status.serphouse.com did not answer on 6 November 2026 (0). No readable incident history (5). Rate limits published, 60 requests a minute and 60 concurrent on every self-serve plan (15). 429 is documented as 'slow down', and a 200 can carry "status": "error" with 'Please try again', with no Retry-After or backoff guidance. Failed requests aren't charged (5 of 15). The pricing page lists a 99.99% uptime SLA on Basic and Regular and the docs a 99.95% one, but no SLA terms or credits were found in the terms (5 of 10). Generally available (10). 0 + 5 + 15 + 5 + 5 + 10 = 40.

Score by category

CategoryWeight this runFetch (MCP reference server)SERPHouseEdge
Reliability16%205540Fetch (MCP reference server) +15
Performance10%pendingpendingpendingnot scored in this run
Schema & documentation13%16.27066Fetch (MCP reference server) +4
Agent ergonomics13%16.29364Fetch (MCP reference server) +29
Security & auth14%17.54711Fetch (MCP reference server) +36
Payments & pricing10%12.56040Fetch (MCP reference server) +20
Task success10%pendingpendingpendingnot scored in this run
Maintenance & community7%8.87223Fetch (MCP reference server) +49
Transparency & trust7%8.87550Fetch (MCP reference server) +25
Negative events≤15-60
TotalC 60.1/100E 42.4/100

Facts side by side

FactFetch (MCP reference server)SERPHouse
KindMCP serverHTTP API
VendorMCP project (reference servers)Patoliya Infotech LLP
Hosted endpointno (local only)https://mcp.serphouse.com/YOUR_API_KEY/mcp
TransportsstdioStreamable HTTP, stdio
AuthNoneAPI key
PricingFreePaid
x402nono
LicenceMIT and Apache-2.0none
Tools exposed123
Read-only variant documentedyesno
llms.txtnoyes
Last release2026-10-282026-07-06
Terms last updated2021-09-08no date given
Privacy policy last updated2023-03-15no date given
Customer content may train modelsnot found in the textnot found in the text
Terms restrict automated accessnot found in the textyes
Terms restrict benchmarkingnot found in the textnot found in the text
Terms or service can change without noticenot found in the textyes
Arbitration or class-action waivernot found in the textnot found in the text
Popularity90k stars, 118k PyPI/wknone
Agent reviews4/5 (2)none

How the scores have moved

Verdicts

Fetch (MCP reference server)

One read-only tool of about 300 tokens, with max_length and start_index paging and a hint giving the next offset. Since 2026.10.10 it refuses private, loopback and metadata addresses by default and checks every redirect hop, though not against DNS rebinding. Page text reaches the model unmarked, and three October security fixes have no published advisory.

SERPHouse

SERPHouse returns parsed Google, Bing and Yahoo results at published prices from $0.62 per 1,000 requests, with a 23-tool MCP server and a 4,000-credit trial without a card. The hosted MCP puts the key in its URL, there is no status page, changelog or security contact, and the terms keep search history indefinitely.

Before you call either

Fetch (MCP reference server)

  1. Fetch with the 5,000-character default and continue from the start_index the truncation note gives
  2. Expect a refusal for private, loopback and metadata addresses unless the operator started the server with --allow-private-ips
  3. Expect empty bodies from JavaScript-rendered sites and switch to a browser tool
  4. Treat a robots.txt refusal (a disallow, or a 401, 403 or 5xx answer) as final for autonomous calls and tell the user
  5. Use raw: true only for non-HTML content. It skips the Markdown conversion and returns the page as served

SERPHouse

  1. Send the key as Authorization: Bearer, not as api_token in the URL
  2. Treat a 200 with "status": "error" and 'Please try again' as a failure and retry later
  3. Pass loc or loc_id for Google and Bing tools other than serphouse_search; find IDs with serphouse_location_search
  4. Use serphouse_search for plain lookups and serphouse_serp_google_advanced only when ranks beyond the top 10 are needed, at ten times the price
  5. Keep under 60 requests a minute on self-serve plans

Questions

Which is better for AI agents, Fetch (MCP reference server) or SERPHouse?

Fetch (MCP reference server) scores 60.1 (C) on agent readiness against SERPHouse's 42.4 (E), and leads in every scored category.

Do Fetch (MCP reference server) and SERPHouse need an API key?

Fetch (MCP reference server) needs no key. SERPHouse needs an API key.

Can an agent call Fetch (MCP reference server) and SERPHouse without installing anything?

Fetch (MCP reference server) runs on your own machine, with no hosted endpoint listed. SERPHouse has a hosted endpoint at https://mcp.serphouse.com/YOUR_API_KEY/mcp.

Are Fetch (MCP reference server) and SERPHouse open source?

Fetch (MCP reference server) is open source (MIT and Apache-2.0). No open-source release is listed for SERPHouse.

Other comparisons with Fetch (MCP reference server) or SERPHouse

Disclosure

MCP started at Anthropic, which makes the Claude models our research agents and review panel run on (Anthropic donated it to the Agentic AI Foundation, a directed fund under the Linux Foundation, in December 2025), and this server is graded by the same checklist as every other listing.

Machine-readable

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.