Find your next open-source project

Search by name, purpose, or keyword — or paste a GitHub repository URL

pugarte7/spanish-top-tech-companies

Productivity

MCP server that gives AI agents Spanish tech job listings with company salary data and lets them submit applications.

600
21
JavaScript
MIT
2 months old

What it does

The project is an MCP server named tech-pay-spain that exposes Spanish tech job data and salary statistics to AI agents. It crawls more than 400 public job portals daily (Greenhouse, Workday, Ashby, Lever, SuccessFactors and others), keeps roles located in Spain or remote roles open to Spain/Europe, and attaches each company's median base salary for that role from Levels.fyi data. Agents can search listings, inspect an application form, and submit applications on the user's behalf after confirmation.

The ranking website at spanishtoptechcompanies.com shows the same dataset for human browsing: companies ranked by median base salary, salaries by role, autonomous community and experience level, and a remote/hybrid/office breakdown.

Connect your agent

No registration, API key or installation is required. Add the Streamable HTTP endpoint as an MCP server in any MCP-capable agent:

https://spanishtoptechcompanies.com/mcp

Users reported working mainly with Codex and Cursor; Claude, Claude Code, Instinct and Hermes are also listed as supported. Alternatively, run the same server over stdio with Node 22 or newer and no dependencies:

node scripts/mcp-stdio.mjs

The README documents an optional Hermes skill install with hermes skills install pugarte7/spanish-top-tech-companies/skills/tech-pay-spain; the skill source is the agent skill file.

Tools available to the agent

The server exposes eight tools: find_jobs for filtered search, job_questions to read a live application form and split questions between you and the agent, application_guide for the apply procedure, report_application for recording outcomes, salary_stats, top_companies, company_salaries and list_filters. The agent selects the tool from natural-language requests.

Applying through the agent

Give the agent your CV and preferences once, then ask it to apply. It searches, drafts the cover letter and answers from your CV, shows you the full application, and submits only after you approve. Sensitive questions such as salary expectations or work permit are batched into one message, and privacy consents are confirmed by you. The website stores nothing about you; the profile and application history stay in the agent's memory or in a file such as ~/job-search.md.

Submitting requires a browser on the agent side; if none is available it opens a headless one (bundled Chromium, Claude in Chrome, or the Playwright MCP server). If that fails, it hands you the link and answers to submit manually. CAPTCHAs and email verification codes are passed back to you or read from a connected mailbox.

Constraints and data notes

Salaries come only from Levels.fyi entries whose location is Spain, and companies need at least 5 salary reports across all countries and roles to be listed; companies with fewer than 3 Spanish salaries are sorted last. Figures are rounded to €1,000, refresh weekly for salaries and daily for jobs, and may be incomplete or outdated. Results are reported salary summaries, not offers or financial/legal advice.

Code is MIT-licensed; data under data/ is CC BY-SA 4.0, and Levels.fyi data remains subject to its own terms. Raw datasets are published as data.json and jobs.json, with a machine-readable summary in llms.txt. The salary methodology is detailed in METHODOLOGY.md.

HotRepo's take

On the 2026-11-18 JavaScript daily list the repository recorded 595 stars, 21 forks and 66 stars gained that day, reaching position 10 in that language ranking. That is a single snapshot, so it supports a same-day observation but not a growth trend. The design choice that stands out technically is the split between a zero-auth hosted MCP endpoint and a dependency-free stdio variant, which lets agents query live job data without any local setup, while the approval gate keeps submission under user control.

Written by AI from the project's README, updated . The project's own docs are the source of truth.

You might also like

  1. strelov1/freehireA self-hostable job search engine that crawls company ATS pages directly, then unifies and deduplicates the postings.Go3,009