Data and privacy

What the server returns

The MCP server returns Estel's market data:

• open tech roles (leads)
• the companies hiring for them
• company news signals
• aggregate statistics

It covers the United Kingdom, Ireland, Germany, Austria, Switzerland, Poland, Romania and Bulgaria.

Lead details come in three levels, matching the Estel app:

Level	Tool	Includes
Card	search_leads	Title, company, location, salary, estimated placement fee, seniority, work and contract type, skills, posting date
Preview	get_lead	Everything on the card, plus the role summary, soft skills, a note about the company, the same role in other cities, and whether the role is still open
Enriched	enrich_lead (1 credit)	Everything in the preview, plus links to the original job postings, experience required, industry, applicant count, job and company tags, and the company's website, size, blue-chip status and social profiles

Company profiles and signals include only company-level information, such as the company's website, size, locations, hiring summary and news.

What the server never returns

Responses are built from fixed schemas. A field that isn't in a schema can't be sent. By design, the schemas leave out:

• Contact and buyer data. No hiring managers, decision makers, contact names, email addresses or phone numbers. Signals summarise public news, so a headline can mention someone the story is about, such as a newly appointed CTO.
• Information about Estel users. Nothing about who else viewed, saved or engaged with a lead, and no hitlists, notes or workflow status.
• Where Estel collected a record. Source platforms aren't named. Job posting links in enriched details are included as they are, so they may point to the site where the job was posted.

What your usage records

• Each tool call: the tool, the outcome, how long it took, how many records it returned, and the names of the arguments used. The argument values aren't logged, so search terms and ids don't appear in logs.
• Each lead you enrich, so it stays unlocked for you and isn't charged again.
• When each API key and connected app was last used, shown in Settings → API & MCP.

Data you send to AI apps

When you connect an AI assistant, the results of its tool calls go to that assistant's provider, such as Anthropic for Claude, Cursor or GitHub. That provider handles the results under its own terms. To limit what an assistant can do, use a read-only API key. To cut off access, disconnect the app or revoke the key in Settings → API & MCP.