Jobicy/remote-jobs-mcp-server

Connect an AI assistant to Jobicy to find public remote jobs, read individual vacancies, discover companies and career roles, and resolve Jobicy filter names. The production MCP service is hosted at https://jobicy.com/mcp. This repository also contains a standalone Node.js MCP server for clients and hosting platforms that run a local process. It uses only public Jobicy API endpoints and does not contain the site's backend.

Connect your MCP client

Streamable HTTP URL: https://jobicy.com/mcp

In an MCP client that accepts a remote server URL, add a server named jobicy and paste that URL. If the client asks for a transport, select Streamable HTTP. No Jobicy account or access token is required for these public, read-only tools.

Clients that use a JSON MCP configuration can use this example:

{
  "mcpServers": {
    "jobicy": {
      "url": "https://jobicy.com/mcp"
    }
  }
}

Older clients that support only SSE can connect to https://jobicy.com/mcp/sse with transport type sse. The client handles the associated /messages route. The MCP discovery document is also available.

After connecting, ask your assistant to list available MCP tools. You should see search_jobs, get_job, and the other tools below.

Run the repository server

Requires Node.js 20 or newer. Install dependencies with pnpm install --frozen-lockfile, then run node server.js. The process speaks MCP over standard input and output; it is intended to be launched by an MCP client or hosting platform. No Jobicy credentials are required. All job, company, role, and taxonomy data comes from the public Jobicy APIs.

A client that supports local MCP commands can use:

{
  "mcpServers": {
    "jobicy": {
      "command": "node",
      "args": ["/absolute/path/to/remote-jobs-mcp-server/server.js"]
    }
  }
}

For Glama's Dockerfile configuration, build the repository with pnpm install --frozen-lockfile and start the local process with mcp-proxy -- node server.js. Do not configure mcp-remote or the hosted /mcp URL as the container command: Glama's repository build runs the MCP process from this source. The live Jobicy endpoint remains available independently.

Find jobs

Use search_jobs for most searches. It returns compact vacancies without the full job description. For example, ask your assistant:

Find up to 10 remote backend engineering jobs open to candidates in the UK. Show the title, company, salary if available, and Jobicy link.

An equivalent tool call is:

{
  "name": "search_jobs",
  "arguments": {
    "query": "backend engineer",
    "geo": "uk",
    "limit": 10
  }
}

The response contains jobs, jobCount, nextCursor, hasMore, and appliedFilters. jobCount counts only jobs in that response, not all matching jobs. To request another page, pass nextCursor back as cursor with the same filters. Treat the cursor as an opaque value; changing filters invalidates it.

For one vacancy's complete description, call get_job with its numeric id. The geo filter describes candidate eligibility: a country search may also include jobs open to its broader region or to candidates anywhere. Salary and logo fields are included only when public data is available.

Available tools

Tool	When to use it	Inputs
search_jobs	Search compact public job results with cursor pagination.	Optional query, geo, industry, jobType, jobLevel, company, postedAfter, salaryMin, salaryMax, currency, limit, cursor.
get_job	Read one full public vacancy.	Required id.
get_similar_jobs	Find a short list of related vacancies.	Required jobId; optional limit.
search_companies	Find a specific public company or a small relevant set. This is a search, not a company directory.	At least one of query, industry, or geo; optional limit (up to 10).
get_company	Read one public company profile.	Required id (numeric ID or slug).
get_company_jobs	Find public jobs linked to one company.	Required companyId; optional limit, cursor.
get_role	Read a Jobicy career role guide.	Required role slug.
get_related_roles	Explore roles related to a career role.	Required role; optional limit.
resolve_taxonomy	Turn a location or industry name into a Jobicy filter slug.	Required type (location or industry) and query.
list_taxonomies	Browse available public filter slugs and labels.	Required type: locations, industries, job_types, or job_levels.

If a filter is obvious, call search_jobs directly. When unsure, use resolve_taxonomy: for example, United States resolves to usa, and Developer can resolve to engineering. If a match is uncertain, the tool returns candidates instead of guessing. Location labels in the new tools are shown without display emoji.

Legacy tools

Existing MCP clients can continue to use:

Tool	Inputs	Compatibility
get_jobs	count, geo, industry, tag; aliases location, category, search, keyword.	Keeps the original full jobs feed and JSON in content.text.
get_taxonomies	type: locations or industries.	Keeps the original JSON in content.text.

New tools provide both structuredContent and readable JSON in content. All tools are read-only. Responses do not provide global job or company totals or Jobicy internal statistics.

Resources

Clients that support MCP Resources can read individual public entities using jobicy://jobs/{id}, jobicy://companies/{id}, and jobicy://roles/{slug}. Public reference lists are available at jobicy://taxonomies/locations and jobicy://taxonomies/industries. Resources supplement the tools; they do not expose an all-jobs or all-companies feed.

Help and security

Open a GitHub issue for connection or documentation problems. Report security concerns privately as described in SECURITY.md.

License

This documentation is available under the MIT License.

Jobicy MCP server