Can an AI agent use Zyte?
Zyte sells web scraping tools and hosting for scrapers.
Rank among the 246 products we drove
- Time to get in
- tied #2
- <1minutes
- Getting in on its own
- Steps on its own
- tied #35
- 3/ 5
- Find to Leave, no person
- Core work done
- #99
- 30%
- 3 of 10 core tasks
Partly. Our agent signed up for Zyte on its own in a browser and used its API, MCP server and CLI.
The record
Level confirmed
Not yet
Level from one drive; a second must agree.
- Got in by
- MCP sign-inUnder a minute
- Got as far as
- Pay: reaches the checkoutA card form for $0.06
Connects by
MCP, API, CLI
Agent payments
None found
On its own · 293 doc pages read
Docs it can act on: llms.txt and an MCP server.
On its own · <1 minute
MCP sign-in.
Under half · 3 of 10 core tasks
Did: deploy a scraper project and run, schedule and monitor jobs; export the scraped items as JSON or CSV. 7 not done, among them: use the main scraping API (fetch, rendering, extraction, actions).
Partly
Reaches the checkout. A card form for $0.06.
On its own
Nothing to cancel. It deletes its account.
- On its own
- Under half
- Partly
To reach L3: Do half of the core work (now 30%).
See how it ranks in the web scraping category.
Own Zyte? 7 things to fix. Get your findings, free
Checked 18 Nov 2026: MCP server asks for a sign-in, llms.txt unchanged since the check before.
Find
How an agent connects
MCP, API, CLI and llms.txt found.
MCP server
29 of 34 testedOfficial server. Remote MCP server at mcp.zyte.com/v1/mcp, MCP (the 28 Jul 2026 spec), 39 tools when driven, OAuth with CIMD or DCR.
34 tools mapped in our drive
33 tools named in its docs
API
49 of 53 tested53 operations mapped in our drive
OpenAPI
UnknownCLI
12 of 24 testedshub 2.18.1 (pip install shub).
24 commands mapped in our drive
SDK
All testedpython-scrapinghub 2.9.0.
21 methods mapped in our drive
llms.txt
YesAgent payments
None foundReached the card form and stopped: Zyte API Standard (PAYG), $0.06 a call. It checks out through Chargebee. No CAPTCHA at checkout. It offers PayPal.
Docs
In partCovers getting started, auth and error codes.
Code examples on only some pages.
Describes each operation only in part: completeness 51 of 100.
2 of 9 documented ways to do the core work were tried, and 2 of them worked.
293 pages read
MCP tools
Read without signing in on 3 Nov 2026
fetch_httpfetch_pageextract_from_httpextract_from_browserzyte_api_domain_pricingzyte_api_usage_statsuser_infoscrapy_cloud_list_projectsscrapy_cloud_get_projectscrapy_cloud_get_project_activityscrapy_cloud_list_spidersscrapy_cloud_run_spiderscrapy_cloud_run_scriptscrapy_cloud_list_jobsscrapy_cloud_get_jobscrapy_cloud_cancel_jobscrapy_cloud_update_job_tagsscrapy_cloud_get_job_logscrapy_cloud_get_job_requestsscrapy_cloud_get_job_itemsscrapy_cloud_get_job_item_statsscrapy_cloud_list_periodic_jobsscrapy_cloud_get_periodic_jobscrapy_cloud_create_periodic_job
9 more tools25 more toolsShow fewer
scrapy_cloud_get_projectscrapy_cloud_get_project_activityscrapy_cloud_list_spidersscrapy_cloud_run_spiderscrapy_cloud_run_scriptscrapy_cloud_list_jobsscrapy_cloud_get_jobscrapy_cloud_cancel_jobscrapy_cloud_update_job_tagsscrapy_cloud_get_job_logscrapy_cloud_get_job_requestsscrapy_cloud_get_job_itemsscrapy_cloud_get_job_item_statsscrapy_cloud_list_periodic_jobsscrapy_cloud_get_periodic_jobscrapy_cloud_create_periodic_jobscrapy_cloud_update_periodic_jobscrapy_cloud_delete_periodic_jobscrapy_cloud_get_project_settingsscrapy_cloud_update_project_settingsscrapy_cloud_list_collectionsscrapy_cloud_get_collection_recordsscrapy_cloud_put_collection_recordscrapy_cloud_delete_collection_recordsscrapy_cloud_delete_collection
Get in
Getting in
1 of 2 ways in worked without a person.
The way that worked
MCP sign-inunder a minute
- An inboxNeeded
- A browserNot needed
- A personNot needed
- Agent on a serverAPI, MCP, CLIPartlyGets in without a browser (mcp sign-in) and does 3 of its 10 core tasks that way. For a full account, it stops at sign-up: it works only in a browser.
- Assistant in its own browserLike ChatGPT agentPartlySigns up, confirms its email and does 3 of its 10 core tasks, then stops: the rest needs a paid plan.
- Signed agentNot checkedNot checked: the robots.txt on app.zyte.com asks crawlers to stay off the sign-up page.
Every way in
| Way in | What happened |
|---|---|
| MCP sign-in | Worked |
| Email and code | Stopped |
- CrawlersThe robots.txt on app.zyte.com asks crawlers to stay off sign-up. Nothing to fix.
Use
What an agent can do with Zyte
3 of 10 core tasks done, and 5 of 18 others.
Core tasks
- done there
- tried, not done
- offered, not tried
| Task | Result | Through |
|---|---|---|
| Deploy a Scrapy spider project to the cloud | Done | MCP: done thereCLI: done there |
| Run/schedule a spider job and monitor it in real time | Done | MCP: done thereCLI: done there |
| Browse and export extracted items/data from a finished job | Done | API: done thereMCP: done there |
| Fetch raw HTTP response body for a URL | Blocked | API: offered, not triedMCP: tried, not done |
| Render a page in a headless browser and return HTML | Blocked | API: offered, not triedMCP: tried, not done |
| Auto-extract structured Product data from a page | Blocked | API: offered, not triedMCP: tried, not done |
| Auto-extract structured data for Article, ArticleList, ArticleNavigation, ForumThread, JobPosting, JobPostingNavigation, PageContent, ProductList, ProductNavigation or SERP | Blocked | API: offered, not triedMCP: tried, not done |
| Define custom AI extraction attributes with a JSON schema | Blocked | API: offered, not triedMCP: tried, not done |
| Submit custom HTTP requests (method, headers, POST body) against the target site | Blocked | API: offered, not triedMCP: tried, not done |
| Run in-page browser actions (type, click, wait, scroll, navigate) before extractionIt stays off until a card is saved, even for the free credit. | Blocked | API: offered, not triedMCP: tried, not done |
Surface audit: 135 documented operations: 113 run, 21 blocked by sandbox or plan limits and 1 left out because it spends money or reaches real people. Of the 113 run, 110 passed and 3 didn't; 5 behaved differently from the docs. By surface, 49 on the API, 29 on the MCP server, 21 in the SDK, 12 in the CLI and 2 elsewhere, such as sign-in and webhooks.
Holds up
83of 100
- Reliability88
- Handles mistakes78
The mean of the two.
Docs match
82of 100
- Docs accuracy96
- Docs completeness51
Docs accuracy counts 70%, completeness 30%.
Surface tested
82of 100
110 of 134 passed: the 135 documented operations less the 1 that spends money or reaches real people
What it can't do, and why
- Use the main scraping API (fetch, rendering, extraction, actions) It stays off until a card is saved, even for the free credit.
- Create or scope API keys without a browser Only the dashboard offers keys; no documented API or tool.
- Create or delete projects and accounts without a browser Only the dashboard offers these; account deletion needs an emailed code.
- Raise Scrapy Cloud job units Needs a paid plan; the error does not say so.
5 of 18 other tasks done
- Monitor spider health/QA with Spidermon integrationDone
- Portable Agent Skills package for any Agent-Skills-compatible coding assistantDone
- Build/run a crawler with the open-source Scrapy frameworkDone
- Estimate/calculate project cost before committingDone
- Receive a monthly usage/analytics reportDone
- Download large files (up to 100MB) with retry/resumePaid plan
- Scale crawl concurrency/capacity on demandPaid plan
- Commission a fully managed custom data feed (Zyte Data)Paid plan
- Receive delivered data on a recurring schedule to a warehouse/endpointPaid plan
- Track delivery quality metrics (accuracy, completeness, freshness, coverage)Paid plan
- Take a screenshot of a rendered pageBlocked
- Request from a specific country/geolocationBlocked
- Persist cookies/session across multiple requestsBlocked
- Attach Zyte API unblocking/rendering to a Scrapy Cloud spider via middlewareBlocked
- Automatic CAPTCHA solving (reCAPTCHA, hCaptcha, etc.)Not tried
- Generate a complete Scrapy project (page objects + tests) from a URL and data description via Claude Code pluginNot tried
- Same spider-generation workflow via Codex CLI pluginNot tried
- Same spider-generation workflow via GitHub Copilot CLI pluginNot tried
- Generate/assist Scrapy extraction code inline in VS Code via Web Scraping Copilot extensionIt stays off until a card is saved, even for the free credit.Not tried
- Join the Zyte user community for support and early accessNot tried
The drive used the free plan (Scrapy Cloud); Zyte API needs a card.
Tips for agent builders
- Pass organizationId as a string to Zyte MCP tools; call user_info first to get it.
- Check user_info's useZapi flag: false means Zyte API calls will be refused until a card is saved.
- Scrapy Cloud works on the free plan: deploy with shub (SHUB_APIKEY env var), then drive jobs through MCP scrapy_cloud_* tools.
- MCP job tools take a 'job' key like 123/1/4 and project tools take an integer 'project_id'.
- Cancelling a job is asynchronous: poll get_job until close_reason is cancelled.
- The MCP OAuth access token lives about 3 minutes; refresh before long sequences.
Pay
Paying
Ways to pay checked 2 Nov 2026.
- ResultPartly. Reaches the checkout. A card form for $0.06.
- CheckoutReached the card form and stopped: Zyte API Standard (PAYG), $0.06 a call. It checks out through Chargebee. No CAPTCHA at checkout. It offers PayPal.
Leave
Leaving
- ResultOn its own. Nothing to cancel. It deletes its account.
- Its accountYes, in a browser, without a human. Account, organization and projects deleted in the browser (Account Settings, email code, Delete); key then refused 403 and MCP token 401. No API route.
Daily checks
Checked every day
What our agent finds on the product's own site each day, without an account.
Sign-up page,
Stopped on the newest check; 17 of 17 days
llms.txt,
Found on the newest check; 17 of 17 days
MCP server,
Asks for a sign-in on the newest check; 4 of 17 days
OpenAPI,
Not on its site on the newest check; 17 of 17 days
Compare
How Zyte compares
In Web scraping
2 of the 5 products we drove in Web scraping are higher; no other has L2. As of 2 Nov 2026.
| Product | Level | Find to leave |
|---|---|---|
| Firecrawl | L3 | Find: On its own, Get in: On its own, Use: On its own, Pay: Partly, Leave: Partly. |
| Browse AI | L3 | Find: Partly, Get in: On its own, Use: Partly, Pay: Partly, Leave: On its own. |
| Apify | L1 | Find: On its own, Get in: Stopped, Use: Not tried, Pay: On its own, Leave: Not reached. |
| ScrapingBee | L1 | Find: Partly, Get in: Stopped, Use: Not reached, Pay: Not reached, Leave: Not reached. |
| Zyte (this page) | L2 | Find: On its own, Get in: On its own, Use: Under half, Pay: Partly, Leave: On its own. |
For the team behind Zyte
Agents get stuck at Use and Pay
7 things to fix for agents: 3 small, 3 medium and 1 large
One fix can cover several findings.
Agent blockers: 4 foundAgent surfaces: 5 issues found
Get your findings, freeWhat a drive covers
We mail a private link to an address at zyte.com.
It opens the two findings that matter most in full, and every other by title. There is nothing to sign in to.
- FindClear
- Get inClear
- UseBlocks agents
- PayBlocks agents
- LeaveClear
Or claim Zyte to see where each issue is, read your safety findings with their fixes and ask for a free drive after a fix.
Want Zyte driven every month? See Watch
On our standard
Best or ahead on 5 of 33 criteria, behind on 3. Each against the products we measured on it.
- Docs an agent can navigateAhead of 128 of 187 others (every product), 59 the same.
- Results sized for a context windowAhead of 44 of 70 others (every product), 26 the same.
- Spend an agent can see and capAhead of 106 of 159 others (every product), 46 the same, 7 higher.
- Truthful annotationsAhead of 50 of 91 others (every product), 41 the same.
- Limits known before they hitLevel with 47 of 90 others (every product), 43 higher.
- llms.txt that worksTied for the lowest result 2 of 4 others (web scraping products) the same, 2 higher.
- A tool list that fitsBehind 71 of 91 others (every product), 20 the same.
- Agents may read docs, API and sign-upBehind 130 of 160 others (every product), 30 the same.
All 33 criteria
- A way in with no personTied for the best result 2 of 4 others (web scraping products) the same, 2 lower.
- Sign-in for agents acting for a personLevel with 83 of 158 others (every product), 75 lower.
- Delete by API, checkedIn the middle 46 of 97 others (every product) lower, 43 the same, 8 higher.
- Prices an agent can readLevel with 117 of 153 others (every product), 36 lower.
- Errors a program can readLevel with 82 of 145 others (every product), 47 lower, 16 higher.
- No person in the buy pathLevel with 114 of 138 others (every product), 24 lower.
- Export before leavingIn the middle 73 of 189 others (every product) lower, 74 the same, 42 higher.
- Fast to the first callLevel with 118 of 135 others (every product), 17 lower.
- Descriptions that say when to use the toolLevel with 56 of 91 others (every product), 22 lower, 13 higher.
- The best way in is built for agentsIn the middle 79 of 196 others (every product) lower, 55 the same, 62 higher.
- Tool names that sort themselvesLevel with 85 of 91 others (every product), 6 lower.
- Terms allow an agent's accountLevel with 179 of 189 others (every product), 10 lower.
- Real status codesLevel with 161 of 168 others (every product), 7 lower.
- Errors say what to do nextLevel with 107 of 145 others (every product), 22 lower, 16 higher.
- Typed, tight input schemasLevel with 59 of 91 others (every product), 17 lower, 15 higher.
- Every way in works for an agentIn the middle 61 of 193 others (every product) lower, 72 the same, 60 higher.
- A path written for agentsIn the middle 48 of 187 others (every product) lower, 91 the same, 48 higher.
- An agent payment rail on the coreThe same as all 4 others (web scraping products).
- WebMCP tools on the web appLevel with 133 of 136 others (every product), 3 higher.
- Structured resultsIn the middle 22 of 78 others (every product) lower, 32 the same, 24 higher.
- A plan an agent can buy by APILevel with 157 of 170 others (every product), 13 higher.
- A request id on every answerIn the middle 42 of 143 others (every product) lower, 47 the same, 54 higher.
- Agents pick the right toolLevel with 72 of 91 others (every product), 19 higher.
- Pages fit one fetchLevel with 118 of 187 others (every product), 69 higher.
- All validation problems at onceLevel with 33 of 57 others (every product), 24 higher.
Where Zyte fits
- Wall
- Card before the first useour agent got in another way
- Topics
- MCP serversFind step
- Docs and llms.txtFind step
- MCP sign-inGet in step
- Sign-upGet in step
- MCP and APIUse step
- Delete the accountLeave step
History
Level by drive
One drive so far, on 2 Nov 2026.
One drive so far.
Read more on this step
Pay is where our agent stopped at Zyte.
- Guide · 18 Nov 2026How AI agents pay for software: x402, MPP and checkoutsCard checkouts, x402 and MPP: what an agent can pay with today, and what most products ask instead.Read
- Finding · 18 Nov 2026How far does an AI agent get in a card checkout?Our agent reached the checkout at 88 of the 190 products it set out to buy from, and the card form at 68.Read
- Guide · 28 Oct 2026Tell an AI agent a feature needs a paid planHow to answer an AI agent at a paid-plan wall: 402 or 403, a Problem Details body with the required plan, and an upgrade link a person can open.Read
Every earlier drive is in the history. The level follows these rules. Data under CC BY 4.0.
Found an error? Tell us on this page or read how corrections work.
Questions people ask about Zyte
- Can an AI agent create a Zyte account on its own?
- Yes, in a browser, without a person. The agent signed up in its own browser at app.zyte.com: email, emailed 6-digit code, password and name, with no bot check; asked for a name, our agent answered. The MCP OAuth sign-in (CIMD, PKCE, one Authorize click) used the same account.
- Does Zyte have an MCP server, and can an agent connect to it?
- Yes. Zyte has an official MCP server. The server is hosted, lists 39 tools, signs an agent in with OAuth and speaks the 28 Jul 2026 MCP spec. A new agent can get in on its own: in the drive of 2 Nov 2026, our agent signed up, verified and got its key without a person. Through the MCP server, our agent tried 17 tasks and did 5.
- How does an AI agent get a Zyte API key?
- In the drive of 2 Nov 2026, our agent got its Zyte API key in a browser, without a person. The agent read the Scrapy Cloud key on the dashboard project page in its browser and then used it without a browser with the API and shub. No documented API, CLI or MCP route creates or scopes keys, and the Zyte API key page redirects to a card checkout. The MCP refresh token was revoked without a browser through the authorization server's revocation endpoint.
- Can an AI agent delete its Zyte account?
- Yes, in a browser, without a human. Account, organization and projects deleted in the browser (Account Settings, email code, Delete); key then refused 403 and MCP token 401. No API route.
- Can an AI agent pay for Zyte?
- Partly. Our agent reached a card form for $0.06.
Cite this page
The page as a reference, in plain text, APA and BibTeX.
Cite this
AgentDrives, “Zyte: AI agent test drive”, 2 November 2026, https://agentdrives.com/tools/zyte. CC BY 4.0.
AgentDrives. (2026, November 2). Zyte: AI agent test drive. https://agentdrives.com/tools/zyte
@misc{agentdrives_zyte_2026,
author = {{AgentDrives}},
title = {Zyte: AI agent test drive},
year = {2026},
month = nov,
url = {https://agentdrives.com/tools/zyte},
note = {Agent test drive, CC BY 4.0},
}Something wrong on this page?
We check it against the drive and correct the page, with the date.