Search SylvaraJob job offers from Claude, ChatGPT, Gemini CLI, Cursor, VS Code and any client that speaks the Model Context Protocol. Public data, read-only, no account required.
The connector exposes three read-only tools. Results list SylvaraJob own postings first (published by SMEs on SylvaraJob, with their full public record and a link to sylvarajob.com) and then external postings that other companies publish on their own career sites through Greenhouse, Lever, Ashby or Teamtailor (title, company, city, date and a link to the source only).
| Tool | What it returns | Arguments |
|---|---|---|
buscar_ofertas | Active offers (external ones only if published in the last 90 days), own first and then external; paginated (max. 20 per page). Text search expands every word with profession and sector synonyms in 8 languages (programador ↔ developer ↔ sviluppatore, camarero ↔ waiter, mozo de almacén ↔ warehouse operative), ignoring case, accents, plurals and gender. With provincia or ciudad, offers in that area come first and remote offers from the same country follow (coincide_zona). | texto, terminos (list: the user's term plus its English translation), provincia, ciudad, pais (ISO-2), modalidad (presencial | remoto | hibrido), origen (todas | propias | externas), pagina, por_pagina, idioma |
ver_oferta | Full public record of an own offer (numeric id) or the public summary of an external one (id «ext-N»). Withdrawn offers return similar ones. | id, idioma |
listar_categorias_y_provincias | Categories, provinces, cities, work modes, contract types and companies with counts (own), plus countries, cities and sources (external). | origen, idioma |
Paste the server URL in your client. No key, no login (the connector has no OAuth yet: Phase 2 will add tools for your own account).
| Claude (web and desktop) | Settings → Connectors → Add custom connector → paste the URL. Works on every plan, including Free (one connector). |
|---|---|
| Claude Code | claude mcp add --transport http sylvarajob https://sylvarajob.com/mcp |
| ChatGPT (developer mode) | Settings → Connectors → Advanced → Developer mode → Create → paste the URL (Plus, Pro, Business, Enterprise and Edu). |
| Cursor | Settings → Tools & MCP → Add, or in .cursor/mcp.json: {"mcpServers":{"sylvarajob":{"url":"https://sylvarajob.com/mcp"}}} |
| VS Code / GitHub Copilot | Command Palette → MCP: Add Server → HTTP → paste the URL (VS Code 1.101 or later). Copilot coding agent: repository settings → MCP configuration with "type":"http" and the URL. |
| OpenAI Codex CLI | codex mcp add sylvarajob --url https://sylvarajob.com/mcp |
| Gemini CLI | In ~/.gemini/settings.json: {"mcpServers":{"sylvarajob":{"httpUrl":"https://sylvarajob.com/mcp"}}} |
| Gemini app (custom apps) | Settings → Connected apps → Add a custom app → paste the URL (currently US-only, personal accounts, English). |
| Gemini Enterprise | Custom MCP server data store with Streamable HTTP and no authentication (admin role required). |
| Perplexity (Pro, Max, Enterprise) | Account settings → Connectors → Custom connector → Remote → paste the URL. |
| Mistral Le Chat | Connectors → Add connector → Custom MCP connector → server URL. |
| Microsoft Copilot Studio | Agent → Tools → Add tool → Model Context Protocol → server URL. |
| Windsurf | Settings → Cascade → MCP servers → Add custom server (serverUrl). |
curl -s -X POST https://sylvarajob.com/mcp \
-H 'Content-Type: application/json' -H 'Accept: application/json' -H 'MCP-Protocol-Version: 2025-11-25' \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"buscar_ofertas","arguments":{"texto":"camarero","por_pagina":5}}}'
MCP 2025-11-25 (also accepts 2025-06-18 and 2025-03-26), Streamable HTTP without sessions: each POST returns JSON (no SSE stream; GET returns 405). Methods: initialize, ping, tools/list, tools/call. Tools carry readOnlyHint, destructiveHint, idempotentHint, openWorldHint, inputSchema and outputSchema, and return text plus structuredContent.
Per IP: 120 requests and 60 tool calls per minute, 2,000 requests per hour (HTTP 429 or JSON-RPC error -32000 with the seconds to wait). Results are cached for 60 seconds. Requests over 64 KB are rejected. Browser clients must come from an allowed origin; server-side clients need no Origin header.
External offers are public postings that companies publish on their own career sites via their applicant tracking systems. SylvaraJob imports title, company, location, date and link every six hours, shows them clearly marked as external and always links to the source: details and applications happen on the company website. Companies can ask for their postings to be removed by writing to sylvarajob@gmail.com.
Searches are processed in memory and not stored. A technical log keeps, for 12 months, the method, tool, duration, status, truncated network (/24 or /48), truncated user agent, the name and version your MCP client declares and the identifiers of the offers returned: never the query text. The AI client you use receives what you ask for. See the connector privacy notice.
Questions, errors or removal requests: sylvarajob@gmail.com