Against research APIs
Research that remembers
Get source-backed answers and ask follow-up questions without starting over. Evidence from earlier workspace runs can be reused.
Web data and research for agents
One integration for web scraping
Fetch pages, track conversations and ads, and get source-backed answers through one API.
- Millions of jobs run
- Published success rates
- MCP, TypeScript & Python
- No charge for blocked, timed-out, or empty results
01 · Install
Start the local MCP server.
Start the local Streamable HTTP MCP server, connect your agent to /mcp, or use an SDK.
MCP server
npx -y @scrapebento/mcpPython
pip install scrapebento$5.54 / 1k
Grounded research on Growth
$0.79 / 1k
Page fetches on the same API
0 credits
Blocked, timed-out, or empty results
02 · Why teams pick it
Priced like infrastructure. Measured in public.
Use one API for everything from raw pages to source-backed research.
Against scraping APIs
One integration instead of three
Fetch pages, track social mentions and ads, build company profiles, and run research with one key, one balance, and one response format.
Across every endpoint
Reliability you can verify
Every endpoint is labeled GA, beta, or experimental and backed by a published success rate. Check live performance before you build on it.
Competitor figures reflect published list pricing at the time of comparison. Scrapebento rates use the Growth plan.
03 · What it does
Four jobs, one integration.
Scrape pages, collect social and ad data, profile companies, or run research — all through the same API.
Research
Source-backed answers
Get structured claims with citations and verification status. Follow-up questions can reuse evidence from earlier workspace runs.
Social and community
Track conversations across the web
Find domain mentions across Reddit, Hacker News, and news. Pull supported posts, profiles, timelines, and LinkedIn employee data.
Brand, company, and ads
Company intelligence in one call
Extract logos, colors, and typography. Pull company profiles, SEO, DNS, technographics, and public ad data.
Scraping
Pages in the format you need
Fetch URLs, sitemaps, feeds, or batches as Markdown, JSON, or screenshots. Bot protection is handled automatically, and blocked or empty results are never charged.
Reliability
Published success rates
View live reliability04 · Reliability
Built for unattended use.
Agents cannot read apologies. Failures are clear, bounded, and never quietly billed.
- Published targets and measured success rates for every endpoint
- Automatic refunds for blocked, timed-out, or empty results
- Stable error codes for programmatic handling
- Inline results for fast calls and jobs for slower ones
- Idempotency keys prevent duplicate billing
- Experimental endpoints stay out of SDKs and agent tools by default
05 · How it works
From signup to first result.
01
Create a key
Sign up without a card. The free plan includes 100 monthly credits.
02
Connect your agent or app
Connect through your deployed Streamable HTTP MCP endpoint, run it locally, or use TypeScript, Python, LangChain, CrewAI, or Vercel AI SDK.
03
Call an endpoint
Track debits and refunds by action, so you always know what each result cost.
06 · Pricing
Simple usage-based pricing.
One credit balance covers scraping, search, social data, ads, and research. Start free, choose a monthly plan, or buy one-time credits whenever you need them.
- 1 creditPage fetch
- 5 creditsPer 10 social mentions
- 3 creditsPer 20 X posts
- 40 creditsPer 10 LinkedIn employees
- 7 creditsGrounded research
07 · FAQ
Common questions.
What can I scrape?
Fetch individual URLs, sitemaps, feeds, or batches as Markdown, JSON, or screenshots. Bot-protected sites automatically use a stronger retrieval path.
How is research different from other research APIs?
Scrapebento can reuse evidence from prior workspace runs and returns a verification status for every claim.
Which endpoints should I depend on?
GA endpoints commit to 99% success over 14 days. Beta endpoints commit to 75%. Experimental endpoints carry no commitment and stay out of SDKs by default. Live results are public on the reliability page.
What happens when a site blocks you?
You receive a stable error code and an automatic credit refund. Blocked, timed-out, and empty results are not charged.
Can I try it without a card?
Yes. The free plan includes 100 credits each month.
Get started
Give your agent the web.
Start scraping and researching in minutes. No card required.