Indeed MCP: Job Listings for Your AI Agent, in Any MCP Client

An Indeed MCP server is a Model Context Protocol server that lets an AI agent such as Claude, ChatGPT or Cursor search Indeed job listings and read them as structured data: title, employer, location, posted pay, job type, the full description and the apply link. Indeed now runs its own server for Claude users, and the best Indeed MCP setup for everything else is Scrape.ly's hosted server, because it works in any MCP client, needs no Indeed account, and returns rows your agent can count, filter and export.

This guide compares the two, connects the Scrape.ly server to each client, lists the Indeed Actors an agent can call, and shows a real run from 26 October 2026: five data analyst jobs in Austin, with employer ratings and company size, in 5.0 seconds for $0.01775. It closes with the edges of Indeed data that an agent should know about before it answers.

Indeed's own MCP server and a scraping server

The first Google result for "indeed mcp" on 26 October 2026 is Indeed's own MCP documentation. The server is "offered in beta", runs remotely over streamable HTTP, and offers four tools: Job Search, Job Detail, Get Resume, which reads "the account holder's experience from their Indeed profile to personalize results", and Get Company Data. Two lines decide whether it fits your project. It "follows the MCP Standard Protocol, but is available only for Claude Connector", and the setup steps end with "Sign in with your Indeed account for first-time use."

That's a good match for a job seeker chatting in Claude. Recruiting analytics, salary research, hiring-signal lead lists and job boards need something else: any client, no personal Indeed login, and the same fields every time so the agent can compare employers or schedule a weekly pull.

How to connect the Scrape.ly MCP server

The Scrape.ly MCP server is hosted at https://mcp.scrape.ly. It uses MCP's streamable HTTP transport and signs you in with OAuth or a Scrape.ly API token. The free plan gives you $5 of usage every month without a card.

Claude Code. Add the server, then run /mcp and choose scrapely to approve it:

claude mcp add --transport http scrapely https://mcp.scrape.ly

Claude Desktop. Add a custom connector under Customize → Connectors with the URL https://mcp.scrape.ly and sign in. It sits next to Indeed's connector if you use both.

Cursor. In ~/.cursor/mcp.json, or .cursor/mcp.json for one project:

{
  "mcpServers": {
    "scrapely": {
      "url": "https://mcp.scrape.ly",
      "headers": { "Authorization": "Bearer YOUR_TOKEN" }
    }
  }
}

Without headers, Cursor opens an OAuth sign-in. Tokens are under Settings → API & Integrations.

ChatGPT, VS Code, Windsurf and other clients. Point them at the same URL over streamable HTTP with OAuth or a Bearer header. Scrape.ly's MCP docs have VS Code's config and a local stdio server.

To show the agent only the Indeed tool, add the Actor to the URL:

https://mcp.scrape.ly?actors=openrole/indeed-scraper

The Indeed Actors your agent can call

Three Indeed Actors in the Scrape.ly Store are maintained by Scrape.ly. All read public listings without a login:

Prices are pay per result with platform usage included, as listed on each Store page on 26 October 2026. The Indeed Scraper returns the title, employer and its Indeed page, location, salary exactly as Indeed prints it, job type, workplace setting, experience level, employer rating and review count, posting date, plain and HTML descriptions and the external apply link. With parseCompanyDetails on, it adds the company's website, description, logo and employee range from the same response, without an extra request. The Indeed scrapers page compares all three with sample output.

What the agent does: four tool calls

1. search-actors with "indeed" returns the three Actors with their prices and input fields.
2. fetch-actor-details reads the chosen Actor's schema and README, so the agent knows it needs position, location and country, and that maxItemsPerSearch sets the size.
3. call-actor starts the run and waits up to 45 seconds. A 10-job search typically finishes in several seconds to about 15, per the README, so it usually comes back done.
4. get-actor-output reads the rows, keeping only the fields the agent asks for. Job descriptions run to thousands of characters, so leaving them out of the first read keeps the context lean.

A real run: data analyst jobs in Austin

The prompt: "Use scrapely to find five data analyst jobs in Austin, TX on Indeed, with pay, job type, experience level, the employer's rating and company size."

The agent's call-actor request:

{
  "actor": "openrole/indeed-scraper",
  "input": {
    "position": "data analyst",
    "location": "Austin, TX",
    "country": "US",
    "maxItemsPerSearch": 5,
    "parseCompanyDetails": true
  },
  "waitSecs": 45,
  "callOptions": { "maxTotalChargeUsd": 0.05 }
}

Our run finished in 5.0 seconds with "Saved 5 jobs from 1 searches." It then read positionName,company,salary,jobType,jobSetting,experienceLevel,rating,reviewsCount,postingDateParsed,companyInfo.companySize:

Company details filled in for the four employers with an Indeed page: Apple and the Texas Department of Transportation at 10,000 or more employees, Ericsson at 5,001 to 10,000, and Austin Community College at 1,001 to 5,000, each with its website.

Three things in this sample are worth building into the agent's instructions. Only two of five listings posted pay, and both were public employers; the agent should report "not listed" rather than guess. The first result came from an employer with no Indeed company page, no rating and no company details, the kind of listing worth flagging for a human check. And the Apple listing was two months old, so for "posted this week" the agent should add Indeed's date filter through a search URL or switch to the Full Actor's fromDays.

Cost: 5 × $0.00355 = $0.01775 at list price, platform usage included. Our account was billed $0.0142 after its plan's Store discount.

Prompts that work well

Tell the agent which Actor, how many jobs, and the cap:

• "Use scrapely to get 100 'product manager' jobs in London on the GB site and give me the share that post a salary."
• "Run openrole/indeed-scraper on https://www.indeed.com/cmp/Google/jobs, 50 jobs, and group them by location."
• "With openrole/indeed-jobs-ppr, find remote data engineer jobs posted in the last 3 days, full-time, 25 rows."
• "Compare data analyst pay in Austin, Denver and Atlanta, 30 jobs each, using only listings that state a salary."
• "Stop every run at $0.25."

The last line becomes callOptions.maxTotalChargeUsd, and the run ends when it reaches the cap. For the bigger picture of collecting job data, see our job scraping guide.

What to watch out for with Indeed data

• About 600 to 750 jobs per search. Indeed stops there even when it reports more, and the run says so in its status. For a full market, split by city, filter or company URL; duplicates across searches are removed by default.
• Salary is text. salary is exactly what Indeed prints, including Indeed's own estimates when it shows them. Parse ranges and units ("a year", "an hour") before you average.
• Missing fields are null, ratings 0. A listing without a field gets null, and an employer with no reviews gets a rating and review count of 0, not an estimate.
• Use postingDateParsed. It holds the posting timestamp; postedAt is null.
• Big jobs take minutes. About 2 to 2.5 minutes per 1,000 jobs, because each page of 100 results is its own request. Past 45 seconds the agent simply polls get-actor-run.
• Recruiters are people. Job descriptions sometimes name a recruiter or include their email. Treat that as personal data, drop it if you don't need it, and keep only employer-level fields for analysis.
• Applications stay private. The Actors read public listings only, never applications, resumes or messages.

Spending control for agents

• Cap each run. maxTotalChargeUsd bounds what a run can cost, and the run stops when it reaches it.
• Scope the key. Create a key for the agent under Settings → API & Integrations → API keys with actors:run and runs:read and a daily charge limit.
• Pay ahead. Runs draw from a prepaid balance, so a scheduled job can't turn into a surprise bill.

Is it legal to scrape Indeed with an AI agent?

The Actors read job listings that any visitor can open without logging in, and Scrape.ly Store Actors collect only publicly available data. Indeed sets its own rules for automated access in its terms and robots.txt, so check how they apply to your project, and use the data for analysis rather than republishing listings wholesale. Names and contact details of recruiters are personal data under the GDPR and the CCPA. Our guide to scraping Indeed covers this in more depth; this is not legal advice.

For a REST endpoint with Python, JavaScript and cURL, see the Indeed API page. The same server covers other job sites through their own Actors: see LinkedIn MCP and our guide to scraping Glassdoor.

FAQ

Does Indeed have an MCP server?

Yes. Indeed offers an MCP server in beta with Job Search, Job Detail, Get Resume and Get Company Data tools. Its docs say it's available only as a Claude connector and that you sign in with your Indeed account. For other clients, or without an Indeed login, use Scrape.ly's server at https://mcp.scrape.ly.

What is the Indeed MCP URL?

Indeed's docs connect its server through Claude's connector directory rather than a URL you paste. Scrape.ly's server URL is https://mcp.scrape.ly; add ?actors=openrole/indeed-scraper to expose only the Indeed tool.

Can I use Indeed MCP with ChatGPT or Cursor?

Indeed's own server is limited to Claude. Scrape.ly's works in ChatGPT, Cursor, VS Code, Windsurf and Claude, because any client that supports remote MCP over streamable HTTP can connect to it.

Does it work with Claude?

Yes. Add https://mcp.scrape.ly as a custom connector in Claude Desktop or with claude mcp add in Claude Code, and Claude can run the Indeed Actors and work with the results.

How much does an Indeed MCP setup cost?

The server is free. The Indeed Scraper costs $3.55 per 1,000 jobs, the Full Actor $3.12 and the Lite Actor $0.025, platform usage included. Our five-job run cost $0.01775 at list price, and the free plan includes $5 every month.

Is there an MCP for LinkedIn or Glassdoor jobs?

The Scrape.ly server reaches both through Store Actors: our LinkedIn MCP guide shows job and company runs, and the Glassdoor scraping guide covers job listings with pay ranges and employee reviews.