Head to head · Scraping · October 2026 research run

fastCRW vs Scrapling

Scrapling scores 68.5 (B) on agent readiness against fastCRW's 66.6 (B), and leads in 2 of 7 scored categories. fastCRW leads on schema & documentation, agent ergonomics and transparency & trust. Both do scraping.

Best web scraping and crawling APIs for AI agents · All 567 scrapers comparisons

Which one, for what

fastCRW B

Best for Agents that want Firecrawl-style scrape, crawl, map, search and extract at lower per-page cost, or a small self-hosted engine.

Ahead on

  • Schema & documentation, 91 against 81
  • Agent ergonomics, 91 against 71
  • Transparency & trust, 60 against 52

Also in its favour

  • A hosted endpoint, with nothing to install
  • Free to start without a card

Watch for

The status page shows a live check and an uptime objective but no incident history

Scrapling B

Best for Agents that need to fetch and drive pages locally, including JavaScript-heavy or bot-protected ones, without a hosted scraping bill.

Ahead on

  • Reliability, 65 against 50
  • Payments & pricing, 90 against 53

Also in its favour

  • No key needed to call it

Watch for

Version 0.5.1 with a Beta classifier. v0.4.15, v0.5 and v0.5.1 each carried breaking changes, the last in a patch release

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.

Payments & pricing Scrapling +37

fastCRW 53

No x402, MPP or L402 (0). Plans are public with credit costs per operation and top-up prices per block of credits, $9 per 1,000 on Hobby down to $299 per 500,000 on Scale, but top-ups need a plan, so between plan-only and per-unit (15 of 20). 500 one-off credits with no card, plus up to 500 from onboarding tasks (20). Hosted keys need a person to sign up in a browser; the free self-hosted engine runs with no account but isn't the graded surface (0). 0 + 15 + 20 + 0 = 35. Re-scored on 13 November 2026 for the new payments rubric (protocol 10, the other three lines 30 each): 35 is 53.

Scrapling 90

Free, self-hosted software under BSD-3-Clause with nothing to buy, read with the self-hosted rule. No payment protocol (0), nothing to price (30), free (30), no signup or key (30). 0 + 30 + 30 + 30 = 90.

Agent ergonomics fastCRW +20

fastCRW 91

11 compact MCP tools, two of them the data-source tools added in 0.41.0. Content is capped near 15,000 characters with truncated: true and a resume offset, and map returns at most 100 URLs by default (20 of 25). Formats, main-content filtering, include and exclude selectors, maxLength, startIndex, limit and crawl depth and page caps (20). Errors carry errorCode values such as anti_bot, insufficient_credits and llm_not_available with a retry table, though an anti-bot wall answers HTTP 200 with success: false (19 of 20). Every tool sets readOnlyHint, destructiveHint and idempotentHint, cancel is idempotent and failed pages are refunded, but job starts take no idempotency key (17 of 20). Only url is required, with official Python and TypeScript SDKs (15). 20 + 20 + 19 + 17 + 15 = 91.

Scrapling 71

17 tools, and v0.5 and v0.5.1 both shortened the descriptions. No toolsets or read-only subset (15). css_selector narrows fetch output, extraction can be Markdown, HTML, text or an ARIA snapshot with search, main_content_only is on by default, and network history is paged with a cursor and a limit of 1 to 200 (18). Failures from bad input, the browser and curl come back as tool errors with the message, which the v0.5 notes say were made clearer. No error codes an agent can branch on (12). Every tool carries annotations, read-only for fetches and lists and destructive for actions, JavaScript and state files. HTTP requests retry three times by default with a one-second wait (17). Only url is required for a fetch, with schema defaults. One language, Python, plus a CLI and Docker (9). 15 + 18 + 12 + 17 + 9 = 71.

Reliability Scrapling +15

fastCRW 50

Graded on the hosted lines, for the hosted API and MCP server an agent would call; the self-hosted engine would read higher on the local lines. fastcrw.com/status is a first-party page with a live health check and a 99.9% monthly uptime objective, but no components with history (10 of 20). It lists no incidents and keeps no history we could read (5). Concurrency per plan is published, 3 on Free to 150 on Scale, while the request-rate guard above it is deliberately unpublished (12 of 15). The rate-limits page documents Retry-After and X-RateLimit headers, gives exponential backoff code, and separates 429 (retry) from 402 (stop). No idempotency keys for crawl, extract or monitor starts (13 of 15). Pricing mentions an SLA only as 'Premier + SLA' on Enterprise, with no terms, and the terms promise no uptime (0). The hosted service is sold on paid plans with no beta label, though the engine is at 0.41.0 (10). 10 + 5 + 12 + 13 + 0 + 10 = 50.

Scrapling 65

Read on the local-software lines, since the MCP server runs on the user's machine over stdio or a self-run HTTP port. Installs from PyPI (scrapling[ai]) or Docker (pyd4vinci/scrapling, ghcr.io/d4vinci/scrapling), with Python 3.10 or later and 3.10 to 3.13 in the classifiers (20). tests.yml runs tox on Python 3.10 to 3.13 on macOS for pushes and pull requests to main and dev, and the repository holds about 1,390 test functions, 19 test files for the MCP tools among them. Run results on GitHub couldn't be read on 18 November, so passing on main is unchecked (15 of 25). Five issues were open on 18 November 2026 against 104,200 stars, the oldest from 24 October 2025, and two of them had a reply. The v0.5.1 notes credit fixes from outside contributors and close reported issues such as #540 (20 of 25). Breaking changes carry a warning and a migration guide in the changelog, but v0.5.1 broke remote browser connections in a patch release (10 of 15). Version 0.5.1 with a Development Status 4 Beta classifier (0). 20 + 15 + 20 + 10 + 0 = 65.

Schema & documentation fastCRW +10

fastCRW 91

OpenAPI 3.1 for 11 paths at version 0.41.0 in docs/openapi.json, checked against the server in CI, and typed JSON Schema inputs on every MCP tool (25). llms.txt (86 lines), llms-full.txt and Markdown sources for every docs page (10). Tool descriptions state purpose and how to chain them (crawl returns a job id to poll, map returns URLs only, extract needs an LLM), and the docs have a choose-an-endpoint page, but few descriptions say when not to use a tool (14 of 20). Inputs are typed with required fields and a renderer enum, though crawl and extract take a free-form JSON Schema object (12 of 15). Examples on every page and an errors page with each HTTP status, errorCode and next step (15). Semver releases with a dated CHANGELOG that flags breaking changes (15). 25 + 10 + 14 + 12 + 15 + 15 = 91.

Scrapling 81

Read from the server source (scrapling/core/ai/server.py) and docs/ai/mcp-server.md in the repository. Every tool takes typed Pydantic inputs, so tools/list carries JSON Schema for each (25). No llms.txt in the repository, and scrapling.readthedocs.io answers 404 for /llms.txt. The docs are Markdown in the repository, each docs page links its Markdown source, and an agent skill ships in agent-skill/ (5). Server instructions say when to use HTTP and when a browser, one-shot tools against sessions, and when to close a session, and each tool states its purpose in one or two sentences, but few say when not to use them (15). Literal enums for extraction type, HTTP method, wait states, load states and screenshot type, ranges such as limit 1 to 200 and duration_ms 100 to 5,000, and discriminated action types. Proxy and cookie fields take free dictionaries (12). The MCP guide has example prompts and setup for each client. Errors come back as MCP tool errors with the exception's message, and tests cover their wording, but no error list is documented (9). A dated changelog for every release from v0.1 (13 October 2024) to v0.5.1 (4 November 2026), and the server reports its version (15). 25 + 5 + 15 + 12 + 9 + 15 = 81.

Transparency & trust fastCRW +8

fastCRW 60

The engine and MCP server are AGPL-3.0 and the SDKs MIT, while the hosted control plane is closed under terms that name no legal entity (25 of 30). The terms say crawled data is handled transiently with zero data retention for approved accounts, but the privacy policy is undated, names no controller, gives no retention periods and mentions no DPA (10 of 30). The CHANGELOG flags breaking changes, deprecated v2 extract routes return a warning on every response, and the October 2026 402 change is dated in the docs, with no standing notice policy (10 of 20). Local data-source telemetry is disclosed with an opt-out. The privacy policy names Stripe, Google Analytics, Google Ads and an X pixel, but infrastructure and search providers are unnamed and no data location is given (8 of 20). 25 + 10 + 10 + 8 = 53.

Scrapling 52

BSD-3-Clause in LICENSE and the PyPI classifiers (30). No telemetry or usage reporting found in the source. Page traffic goes to the target sites, with optional DNS-over-HTTPS through Cloudflare. There is no privacy statement for the package (15). No deprecation policy. Breaking changes are announced in the release notes with migration guides for v0.5 and the remote-session change (8). Nothing is collected that we could find, but there is no explicit statement on telemetry (12). 30 + 15 + 8 + 12 = 65.

Score by category

CategoryWeightfastCRWScraplingEdge
Reliability16%5065Scrapling +15
Performance10%pendingpendingpendingnot scored in this run
Schema & documentation13%9181fastCRW +10
Agent ergonomics13%9171fastCRW +20
Security & auth14%4039fastCRW +1
Payments & pricing10%5390Scrapling +37
Task success10%pendingpendingpendingnot scored in this run
Maintenance & community7%9393even
Transparency & trust7%6052fastCRW +8
Negative events≤1500
TotalB 66.6/100B 68.5/100

Facts side by side

FactfastCRWScrapling
KindHTTP APIMCP server
VendorNanofy OÜKarim Shoair
Hosted endpointhttps://fastcrw.com/mcpno (local only)
Transportsstdio, Streamable HTTPstdio, Streamable HTTP
AuthOAuth or keyNone
PricingFreemiumFree
x402nono
LicenceAGPL-3.0BSD-3-Clause
Tools exposed1117
Read-only variant documentednono
llms.txtyesno
MCP registryio.github.fastcrw/crwio.github.D4Vinci/Scrapling
Last release2026-11-042026-11-04
Terms last updated2026-04-04no document linked
Privacy policy last updatedno date givenno document linked
Customer content may train modelsnot found in the textnot read yet
Terms restrict automated accessyesnot read yet
Terms restrict benchmarkingyesnot read yet
Terms or service can change without noticenot found in the textnot read yet
Arbitration or class-action waivernot found in the textnot read yet
Popularity1.2k stars, 881 npm/wk104k stars

How the scores have moved

Verdicts

fastCRW

Agents get 11 compact, annotated MCP tools with bounded output, a public OpenAPI spec, refunds on failed pages and a free self-hosted engine. The hosted service has no status history, publishes no SLA terms, names its operator only on the support page, and documents an MCP address with the API key in the URL path.

Scrapling

Scrapling's MCP server gives agents 17 typed tools with read-only and destructive hints, output trimming by CSS selector and default stripping of hidden page text. It is pre-1.0 (0.5.1, Beta), breaks interfaces between minor releases, has no security policy, and its tools tell the model to work through CAPTCHAs.

Before you call either

fastCRW

  1. Connect to https://fastcrw.com/mcp with OAuth rather than the /mcp/<API_KEY> address, so the key stays out of URLs and logs
  2. Branch on errorCode. Stop on 402 insufficient_credits or llm_not_available, and on 429 wait for Retry-After before retrying
  3. Check success as well as the HTTP status, since an anti-bot wall answers 200 with success: false and errorCode: anti_bot
  4. Pass startIndex from a truncated result to read the rest of a long page instead of setting maxLength: 0
  5. Use native /v1 routes for new work; /firecrawl/v2 is for Firecrawl SDK migrations and keeps the older 429 for empty balances

Scrapling

  1. Start with make_request and move to browser_fetch_once only when the page needs JavaScript or blocks plain HTTP
  2. Pass css_selector on fetch tools to cut output. Snapshots need one match or no selector
  3. Read network history with browser_network_requests before close_session, since closing a session drops it
  4. Check the target site's terms and robots.txt yourself. The MCP tools don't read robots.txt, only the spiders do
  5. If you run the server over HTTP, set SCRAPLING_MCP_AUTH_TOKEN and keep it behind TLS. The token travels in clear over plain HTTP

Questions

Which is better for AI agents, fastCRW or Scrapling?

Scrapling scores 68.5 (B) on agent readiness against fastCRW's 66.6 (B), and leads in 2 of 7 scored categories. fastCRW leads on schema & documentation, agent ergonomics and transparency & trust.

Do fastCRW and Scrapling need an API key?

fastCRW takes an API key or an OAuth sign-in. Scrapling needs no key.

Can an agent call fastCRW and Scrapling without installing anything?

fastCRW has a hosted endpoint at https://fastcrw.com/mcp. Scrapling runs on your own machine, with no hosted endpoint listed.

Are fastCRW and Scrapling open source?

Yes. fastCRW is open source (AGPL-3.0). Scrapling is open source (BSD-3-Clause).

Other comparisons with fastCRW or Scrapling

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.