ChatGPT Scraper MCP Server

Q: Does every answer include sources or citations?

No. Check conversation.usedWebSearch and the available conversation.sources. These records describe pages ChatGPT consulted, not a guaranteed citation map for every sentence. The request has no separate switch to force web search.

Q: Can I choose a model, continue a conversation or read my chat history?

The request accepts prompt and optional timezone. It does not expose model selection, private chat history or conversation continuation. Each call starts a fresh anonymous conversation. A returned model label describes that response, not a model you selected.

Q: What should my agent do if a request fails?

Check for a successful API status, a non-empty conversation.answer and conversation.complete before using a capture. Treat an error response as a failed request, not an empty answer. Failed requests consume no credits. Verify the sources before treating an answer as fact.

Connect your AI agents to HasData's ChatGPT scraper MCP server. Send a prompt and collect the answer, available web sources and model metadata as JSON over Streamable HTTP.

https://mcp.hasdata.com/mcp?apis=chatgpt

Transport: Streamable HTTP
Auth: OAuth or API key
Tools: 1 ChatGPT tool
Price: 10 credits / call

Get API Key Connect your client 1,000 free credits. No credit card required.

Data teams at

Playground

See the answer and the sources behind it

Explore recorded examples of recommendation research and source collection. The agent turns the response into a summary, while the JSON tab shows an abridged tool result.

Ask ChatGPT which uptime tools it recommends for a small team, and list the brands.

ChatGPT · 1 tool Enter to send

Ask ChatGPT which uptime tools it recommends for a small team, and list the brands.

hasdata_chatgpt_chat_getChatgptAnswer · 200 OK sample call

For a small team, I'd start with UptimeRobot for basic uptime checks or Checkly if you need to test API and browser flows.

Use UptimeRobot for URL uptime checks and downtime alerts.

Checkly combines uptime checks with API monitoring and Playwright-based browser tests. Choose it when a working homepage is not enough and you need to test an actual user flow.

Pingdom adds website-performance monitoring to uptime checks, making it a fit when you want to track speed as well as availability.

Uptime Kuma is the open-source, self-hosted option. You control the monitoring setup, but also run and maintain the monitoring server.

Better Stack connects monitoring with on-call, incident management and status pages. It is useful when the team needs a response workflow, not just an alert.

StatusCake covers uptime and SSL monitoring. It is another option to compare for straightforward website checks.

Sources

UptimeRobot · https://uptimerobot.com/website-monitoring/

Checkly · https://www.checklyhq.com/

Uptime Kuma · https://github.com/louislam/uptime-kuma

hasdata_chatgpt_chat_getChatgptAnswer {"prompt":"Which tools can a small team use to monitor website uptime? Include sources.","timezone":"Europe/Berlin"}

{

"url": "https://api.hasdata.com/scrape/chatgpt/chat/?prompt=Which%20tools%20can%20a%20small%20team%20use%20to%20monitor%20website%20uptime%3F%20Include%20sources.&timezone=Europe%2FBerlin",

"status": 200,

"json": {

"conversation": {

"answer": "For a small team, I’d focus on tools that are easy to set up, have useful alerting, and don’t require adopting a full observability platform.\n\n| Tool | Best for | Small-team highlights |\n|---|---|---|\n| **UptimeRobot** | **Best overall / budget** | Free tier supports 50 monitors with 5-minute checks; HTTP, ping, port, keyword, API, DNS, SSL and other monitoring; alerts via email and integrations. Paid Team plan adds 30-second checks, 3 seats, status pages and webhooks. |\n| **Checkly** | Developer-focused teams | Combines uptime monitoring with API and browser/Playwright synthetic tests. Free tier has 10 uptime monitors and 1,000 browser checks; paid plans add more monitors and locations. |\n| **Pingdom** | Uptime + performance | Checks from 100+ worldwide servers, supports alerts and public status pages, and adds root-cause analysis and performance monitoring. There's a 30-day free trial. |\n| **Uptime Kuma** | Self-hosted / open source | Free and open source, with no SaaS subscription. Good for a technical team that wants control over its monitoring infrastructure, although you have to operate the monitoring server yourself. |\n| **Better Stack** | Monitoring + incident response | Particularly useful if you want uptime monitoring integrated with on-call, incident management, and status pages rather than just basic downtime alerts. |\n| **StatusCake** | Affordable broad monitoring | Offers uptime, SSL and other website monitoring, with a free tier and paid monitoring starting at relatively low cost. |\n\n### My picks\n\n- **Just need to know when the site goes down:** **UptimeRobot**. Its free 50-monitor allowance is unusually generous for a small team.\n- **You have developers and want to test actual user/API flows:** **Checkly**.\n- **You want uptime plus incident/on-call tooling:** **Better Stack**.\n- **You want to self-host:** **Uptime Kuma**.\n- **You care about website performance as well as availability:** **Pingdom**.\n\nFor most small teams, I'd start with **UptimeRobot** and configure at least an HTTP check, SSL/domain-expiration monitoring, and alerts to the team's existing chat/incident channel. You can move to Checkly or Better Stack if simple \"is the URL responding?\" checks aren't enough.\n\nUptimeRobot's 2026 comparison of uptime-monitoring tools is also a useful current comparison of these categories.",

"title": "Uptime Monitoring Tools",

"model": "gpt-5-6",

"usedWebSearch": true,

"complete": true,

"finishReason": "stop",

"sources": [

{

"title": "Who Should Use UptimeRobot's Free Plan? | UptimeRobot Help Center",

"url": "https://help.uptimerobot.com/en/articles/11604710-who-should-use-uptimerobot-s-free-plan",

"domain": "help.uptimerobot.com"

{

"title": "Understanding UptimeRobot Monitor Types: A Guide to Essential Services | UptimeRobot Help Center",

"url": "https://help.uptimerobot.com/en/articles/11358441-understanding-uptimerobot-monitor-types-a-guide-to-essential-services",

"domain": "help.uptimerobot.com",

"publishedDate": "2026-07-31T00:00:00.000Z"

{

"title": "Checkly Pricing Plans: Flexible Synthetic Monitoring Solution",

"url": "https://www.checklyhq.com/pricing/",

"domain": "www.checklyhq.com"

}

]

}

Ask ChatGPT which uptime tools it recommends for a small team, and list the brands.

hasdata_chatgpt_chat_getChatgptAnswer · 200 OK sample call

For a small team, I'd start with UptimeRobot for basic uptime checks or Checkly if you need to test API and browser flows.

Use UptimeRobot for URL uptime checks and downtime alerts.

Checkly combines uptime checks with API monitoring and Playwright-based browser tests. Choose it when a working homepage is not enough and you need to test an actual user flow.

Pingdom adds website-performance monitoring to uptime checks, making it a fit when you want to track speed as well as availability.

Uptime Kuma is the open-source, self-hosted option. You control the monitoring setup, but also run and maintain the monitoring server.

Better Stack connects monitoring with on-call, incident management and status pages. It is useful when the team needs a response workflow, not just an alert.

StatusCake covers uptime and SSL monitoring. It is another option to compare for straightforward website checks.

Sources

UptimeRobot · https://uptimerobot.com/website-monitoring/

Checkly · https://www.checklyhq.com/

Uptime Kuma · https://github.com/louislam/uptime-kuma

hasdata_chatgpt_chat_getChatgptAnswer {"prompt":"Which tools can a small team use to monitor website uptime? Include sources.","timezone":"Europe/Berlin"}

{

"url": "https://api.hasdata.com/scrape/chatgpt/chat/?prompt=Which%20tools%20can%20a%20small%20team%20use%20to%20monitor%20website%20uptime%3F%20Include%20sources.&timezone=Europe%2FBerlin",

"status": 200,

"json": {

"conversation": {

"title": "Uptime Monitoring Tools",

"model": "gpt-5-6",

"usedWebSearch": true,

"complete": true,

"finishReason": "stop",

"sources": [

{

"title": "Who Should Use UptimeRobot's Free Plan? | UptimeRobot Help Center",

"url": "https://help.uptimerobot.com/en/articles/11604710-who-should-use-uptimerobot-s-free-plan",

"domain": "help.uptimerobot.com"

{

"title": "Understanding UptimeRobot Monitor Types: A Guide to Essential Services | UptimeRobot Help Center",

"url": "https://help.uptimerobot.com/en/articles/11358441-understanding-uptimerobot-monitor-types-a-guide-to-essential-services",

"domain": "help.uptimerobot.com",

"publishedDate": "2026-07-31T00:00:00.000Z"

{

"title": "Checkly Pricing Plans: Flexible Synthetic Monitoring Solution",

"url": "https://www.checklyhq.com/pricing/",

"domain": "www.checklyhq.com"

}

]

}

Ask ChatGPT what MCP is, and show the web sources it searched.

hasdata_chatgpt_chat_getChatgptAnswer · 200 OK sample call

MCP (Model Context Protocol) is an open standard that connects AI applications to external data sources, tools and services through a common interface.

An MCP client can discover and call tools from different servers using the same protocol, instead of requiring a custom integration for every service.

Sources

Official MCP documentation · https://modelcontextprotocol.io/

MCP Registry reference · https://registry.modelcontextprotocol.io/docs

TypeScript SDK · https://ts.sdk.modelcontextprotocol.io/v2/

hasdata_chatgpt_chat_getChatgptAnswer {"prompt":"In two sentences, explain what Model Context Protocol (MCP) is. Cite the official modelcontextprotocol.io documentation.","timezone":"Europe/Minsk"}

{

"url": "https://api.hasdata.com/scrape/chatgpt/chat/?prompt=In%20two%20sentences%2C%20explain%20what%20Model%20Context%20Protocol%20(MCP)%20is.%20Cite%20the%20official%20modelcontextprotocol.io%20documentation.&timezone=Europe%2FMinsk",

"status": 200,

"json": {

"conversation": {

"answer": "The **Model Context Protocol (MCP)** is an open standard that enables AI applications to connect to external data sources, tools, and services through a common interface. It allows AI models to access information and interact with external systems consistently, without requiring custom integrations for every service.\n\n**Official documentation:** Model Context Protocol (MCP)",

"title": "Explain MCP Docs",

"sources": [

{

"title": "Official MCP Registry Reference",

"url": "https://registry.modelcontextprotocol.io/docs",

"domain": "registry.modelcontextprotocol.io"

{

"title": "MCP TypeScript SDK",

"url": "https://ts.sdk.modelcontextprotocol.io/v2/",

"domain": "ts.sdk.modelcontextprotocol.io"

{

"title": "MCP TypeScript SDK | MCP TypeScript SDK (v1)",

"url": "https://ts.sdk.modelcontextprotocol.io/",

"domain": "ts.sdk.modelcontextprotocol.io"

}

"model": "gpt-6",

"usedWebSearch": true,

"complete": true,

"finishReason": "stop"

}

Tools

What the ChatGPT scraper returns

1 tool

Send a question in prompt and receive the answer as Markdown inside conversation. When ChatGPT searches the web, inspect its sources and usedWebSearch flag. Your agent can extract brand mentions or compare recommendations from the answer, without treating generated text as verified fact.

+ hasdata_chatgpt_chat_getChatgptAnswer Capture an anonymous ChatGPT answer and its available web sources 2 params 10 credits

Try

Parameter reference ↗

Parameters of hasdata_chatgpt_chat_getChatgptAnswer
Parameter	Type	Required	Example
prompt	string	required	"Which tools can a small team use to monitor website uptime? Include sources."
timezone	string	optional	"Europe/Berlin"

response

conversation.answer
conversation.sources
conversation.sources[].url
conversation.sources[].domain
conversation.title
conversation.model
conversation.usedWebSearch
conversation.complete

Base calls cost 10 credits. Failed calls are not billed.

Connect

Connect Claude, Cursor or your MCP client

Use the hosted URL with a HasData API key, or sign in through OAuth where your client supports it. No OpenAI API key or ChatGPT login is required.

Claude Code

zsh — hasdata-chatgpt

$claude mcp add --transport http hasdata-chatgpt \

> "https://mcp.hasdata.com/mcp?apis=chatgpt" \

> --header "x-api-key: YOUR_API_KEY"

· One line, no restart. Run /mcp to confirm 1 tool.

Check the connectionIn the session

/mcp

hasdata-chatgpt · connected · 1 tool

Cursor

~/.cursor/mcp.json

{

"mcpServers": { "hasdata-chatgpt": {

"url": "https://mcp.hasdata.com/mcp?apis=chatgpt",

"headers": { "x-api-key": "YOUR_API_KEY" }

} }

}

· Add this entry to ~/.cursor/mcp.json. Servers already in the file stay as they are.

· Project-level config also works at .cursor/mcp.json

Check the connectionAfter saving the file

Settings → MCP

hasdata-chatgpt · 1 tool enabled

Claude Desktop

Recommended · OAuth connector

1.Settings → Connectors → Add custom connector

2.Paste https://mcp.hasdata.com/mcp?apis=chatgpt

3.Sign in with HasData

~/Library/Application Support/Claude/claude_desktop_config.json

{

"mcpServers": { "hasdata-chatgpt": {

"command": "npx",

"args": ["-y", "mcp-remote", "https://mcp.hasdata.com/mcp?apis=chatgpt",

"--header", "x-api-key:YOUR_API_KEY"]

} }

}

· Add this entry to ~/Library/Application Support/Claude/claude_desktop_config.json. Servers already in the file stay as they are.

· The config file only takes stdio servers, so a static key rides the mcp-remote bridge. On Windows the file lives at %APPDATA%\Claude\claude_desktop_config.json.

Check the connectionReopen the app

Settings → Connectors

hasdata-chatgpt · connected · 1 tool

VS Code / Copilot

.vscode/mcp.json

{

"servers": { "hasdata-chatgpt": {

"type": "http",

"url": "https://mcp.hasdata.com/mcp?apis=chatgpt",

"headers": { "x-api-key": "YOUR_API_KEY" }

} }

}

· Add this entry to .vscode/mcp.json. Servers already in the file stay as they are.

· Agent mode picks the server up on save.

Check the connectionAgent mode

Ctrl/⌘ + Shift + P → MCP: List Servers

hasdata-chatgpt · running · 1 tool

Windsurf

~/.codeium/windsurf/mcp_config.json

{

"mcpServers": { "hasdata-chatgpt": {

"serverUrl": "https://mcp.hasdata.com/mcp?apis=chatgpt",

"headers": { "x-api-key": "YOUR_API_KEY" }

} }

}

· Add this entry to ~/.codeium/windsurf/mcp_config.json. Servers already in the file stay as they are.

· Windsurf reads serverUrl, not url. Header auth only.

Check the connectionIn Cascade

Cascade → MCP servers

hasdata-chatgpt · 1 tool

Cline

cline_mcp_settings.json

{

"mcpServers": { "hasdata-chatgpt": {

"type": "streamableHttp",

"url": "https://mcp.hasdata.com/mcp?apis=chatgpt",

"headers": { "x-api-key": "YOUR_API_KEY" }

} }

}

· Add this entry to cline_mcp_settings.json. Servers already in the file stay as they are.

· Header auth only. The type field tells Cline this is a remote server, not a command.

Check the connectionIn the MCP panel

MCP Servers → Installed

hasdata-chatgpt · 1 tool

Codex CLI

~/.codex/config.toml

[mcp_servers.hasdata-chatgpt]

url = "https://mcp.hasdata.com/mcp?apis=chatgpt"

env_http_headers = { "x-api-key" = "HASDATA_API_KEY" }

· Add this entry to ~/.codex/config.toml. Servers already in the file stay as they are.

· TOML, not JSON. Export HASDATA_API_KEY in the shell that starts Codex, then open a new session and use /mcp to inspect its tools.

Check the configurationConfigured servers

codex mcp list

Confirm hasdata-chatgpt, its URL and enabled status.

Gemini CLI

~/.gemini/settings.json

{

"mcpServers": { "hasdata-chatgpt": {

"httpUrl": "https://mcp.hasdata.com/mcp?apis=chatgpt",

"headers": { "x-api-key": "YOUR_API_KEY" }

} }

}

· Add this entry to ~/.gemini/settings.json. Servers already in the file stay as they are.

· httpUrl, not url: Gemini keeps url for SSE servers. Header auth only.

Check the connectionIn the session

/mcp

hasdata-chatgpt · 1 tool

Raw HTTP / curl

zsh — hasdata-chatgpt

$curl -N "https://mcp.hasdata.com/mcp?apis=chatgpt" \

> -H "x-api-key: YOUR_API_KEY" \

> -H "Content-Type: application/json" \

> -H "Accept: application/json, text/event-stream" \

> -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'

· Fallback for any client not listed. Streamable HTTP wants both Accept types, and leaving one out returns 406.

Check the connectionCall it yourself

curl … tools/list

{ "tools": [ … 1 item ] }

Scope

Compare ChatGPT answers with search results

Add Google Search to inspect the pages alongside a ChatGPT answer. Your client receives only the tools for the selected services, using the same HasData account.

ChatGPT 1 tool See the full MCP server →

https://mcp.hasdata.com/mcp?apis=chatgpt,google_serp

9active tools in your server

Build vs Buy

Free code, paid in hours

Getting an answer is only the start. Someone still has to keep sessions, response capture and parsing working. HasData handles that collection work for your ChatGPT research.

Self-hosted costs and hours are illustrative budgets for one custom scraper. Engineering time is separate from infrastructure spend.

Scenario

A script I run once

Your own scraperexample budget

$0runs locally

+ 3–6 hsetup once

Adapt a script and run it3–6 h once

Parser upkeepNo recurring work

Request handlingBasic retries

Proxy and hosting$0, local run

A one-off run on your machine. Keep full control over field selection.

mcp.hasdata.comFree tier

$0/ month

Includedparser maintenance

Setup one line

Proxies and source-side retries Included

Concurrent requests 5

25 calls at 10 credits 250 of 1,000 credits

Free credits renew every month. No credit card.

leans self-hosted

Keep the free script

If a working scraper already gets the result, keep it. Use HasData when the job becomes recurring.

A weekly report

Your own scraperexample budget

$0–40/ month

+ 2–4 h / moupkeep

Setup3–6 h once

Parser fixes1–2 h / mo

Run checks and retries1–2 h / mo

Proxy and hosting$0–40 / mo

A scheduled report with occasional parser fixes and checks when a run fails.

mcp.hasdata.comFree tier

$0/ month

Includedparser maintenance

Setup one line

Proxies and source-side retries Included

Concurrent requests 5

100 calls at 10 credits 1,000 of 1,000 credits

Keep your report schedule and storage. HasData runs the scraper.

leans hosted

Skip the upkeep

A free script still needs an owner. Use HasData to collect the data without making parser fixes part of your reporting routine.

A pipeline in production

Your own scraperexample budget

$40–120/ month

+ 6–10 h / moupkeep

Setup3–6 h once

Parser fixes2–4 h / mo

Recovery and monitoring4–6 h / mo

Proxy and hosting$40–120 / mo

Regular collection through proxies, with time budgeted for failed runs and source changes.

mcp.hasdata.comStartup, billed yearly

$49/ month

Includedparser maintenance

Setup one line

Proxies and source-side retries Included

Concurrent requests 5

5,000 calls at 10 credits 50,000 of 200,000 credits

Hosted extraction, with headroom left in the plan for other calls.

leans hosted

Use the hosted server

A pipeline needs data every run. Use HasData instead of owning proxy setup, parser fixes and source-side retries.

A feature in my product

Your own scraperexample budget

$300+/ month

+ On-callplus parser upkeep

Setup3–6 h once

Parser fixes2–4 h / mo

Extraction incidentsYour team responds

Proxy and hosting$300+ / mo

A customer-facing feature with a larger proxy budget and extraction incidents to support.

mcp.hasdata.comBasic, billed yearly

$99/ month

Includedparser maintenance

Setup one line

Proxies and source-side retries Included

Concurrent requests 15

50,000 calls at 10 credits 500,000 of 1,000,000 credits

Your app still owns field mapping and error handling. The scraper is on us.

leans hosted

Ship the feature

Put engineering time into the feature customers use. HasData maintains the extraction behind it.

Pricing

Plans for your prompt workflow

Calls use your HasData account balance. Choose a plan for the request volume your workflow needs, and use the same credits across MCP and REST.

Free

$0 /mo

Free forever

100 tool calls / month

1,000 credits / month

5 concurrent requests

Every MCP tool on every plan

Failed calls are not billed

One balance across MCP and REST

Community support

Start free

Startup

$49 /mo

$2.46 / 1k tool calls

20K tool calls / month

Billed annually Save $118

200K credits / month

5 concurrent requests

Every MCP tool on every plan

Failed calls are not billed

One balance across MCP and REST

Email support

Get started

Basic

Recommended

$99 /mo

$0.99 / 1k tool calls

100K tool calls / month

Billed annually Save $238

1M credits / month

15 concurrent requests

Every MCP tool on every plan

Failed calls are not billed

One balance across MCP and REST

Priority email support

Get started

Growth

$208 /mo

$0.69 / 1k tool calls

Billed annually Save $498

3M credits / month

50 concurrent requests

Every MCP tool on every plan

Failed calls are not billed

One balance across MCP and REST

Dedicated account manager

Get started

Monthly tool call volume

Free 20K 100K 300K 500K 1M 2M

Best fit

Basic

tool calls / mo

100K

Concurrency

$ / 1k tool calls

$0.99

Billed annually

$99 /mo

Get Started

Enterprise

Custom price based on required volume

Past 20M credits a month, or terms the self-serve plans do not cover. We shape the contract around your workload. Past 20M credits a month, or need terms the self-serve plans do not cover? We shape the contract, concurrency, and support around your workload.

Credits rollover

Unused credits carry into the next billing period.

Concurrency 2000+

Parallel request limits set to your peak load.

#1 request priority

Highest speed, always first in the queue.

Personal manager

A direct line to the founding team.

SSO

SAML single sign-on for the whole team.

Security review

Security questionnaire, DPA, and controls overview.

Talk to sales Quote within one business day

Use the same scraper through REST

▤ ChatGPT Scraper APISend prompts from your application and receive answers and available web sources as JSON. Scraper API

FAQ

Before you connect

Is this a server for adding tools to ChatGPT?

No. This server retrieves ChatGPT answers for an MCP client such as Claude or Cursor. To use HasData's web-data tools inside ChatGPT, choose the ChatGPT integration instead.

Is this the official OpenAI API?

No. HasData provides an independent ChatGPT scraper that sends an anonymous prompt and captures the answer. It is not a drop-in replacement for OpenAI's developer API, and it does not require your OpenAI API key or ChatGPT login.

Can I use it for brand mentions and AI visibility research?

Yes. Your agent can inspect the returned answer for brand mentions and recommendations, then group the available source URLs by domain. Save captures in your own workflow if you want comparisons over time. Scheduling, scores and alerts are not built into this tool.

Does every answer include sources or citations?

No. Check conversation.usedWebSearch and the available conversation.sources. These records describe pages ChatGPT consulted, not a guaranteed citation map for every sentence. The request has no separate switch to force web search.

Can I choose a model, continue a conversation or read my chat history?

The request accepts prompt and optional timezone. It does not expose model selection, private chat history or conversation continuation. Each call starts a fresh anonymous conversation. A returned model label describes that response, not a model you selected.

How do I set the answer language or timezone?

Write the prompt in the desired language, or ask for that language in the prompt. An IANA timezone, such as Europe/Berlin, gives date-sensitive wording like today or this week a frame of reference. It is not geographic targeting.

Do I need to run a browser or local server?

No. Connect to the hosted Streamable HTTP endpoint using your HasData account. The client configurations above cover supported formats. Your agent sends tool arguments, and HasData handles the request and parsing.

What should my agent do if a request fails?

Check for a successful API status, a non-empty conversation.answer and conversation.complete before using a capture. Treat an error response as a failed request, not an empty answer. Failed requests consume no credits. Verify the sources before treating an answer as fact.

What happens when my credits run out?

Requests stop when your balance cannot cover the next call. HasData does not charge overages. Wait for the next billing cycle or renew your paid subscription early to restore credits.

Is HasData affiliated with OpenAI?

No. HasData is an independent service and is not affiliated with, endorsed by, or sponsored by OpenAI. ChatGPT is a trademark of its respective owner.