remotive-mcp-server
MCP server that provides tools to search and retrieve remote jobs from Remotive's public job board, with no API key required.
README
remotive-mcp-server
An MCP server that gives Claude (or any MCP client) two tools for searching Remotive's public remote-job board: remotive_search_jobs and remotive_get_job. No API key or signup required — Remotive's job endpoint is free and public.
Built as a small, focused demo of MCP server design: real API integration, defensive handling of an upstream quirk, Pydantic-validated inputs, and a verifiable evaluation set.
What it does
remotive_search_jobs— search current Remotive listings by keyword and/or category, returns a compact markdown or JSON digest (title, company, category, tags, location, salary, post date, apply URL).remotive_get_job— given a job id from a search result, returns the full listing including the plain-text job description.
Example prompts once connected to Claude:
- "Find remote Rails jobs on Remotive and summarize the top 3."
- "Pull the full description for job 2069746."
- "What DevOps roles are open right now, and what do they pay?"
A real-world gotcha this server works around
Remotive's public /api/remote-jobs endpoint sits behind a Cloudflare cache that, in testing, keyed its cache only on the request path — not the query string. Every request, regardless of search, category, or limit params, returned an identical cached response (confirmed via the cf-cache-status: HIT response header and byte-identical results across differing params).
To make filtering actually work for callers, this server fetches the full current listing and applies search/category filtering client-side rather than trusting the upstream query params. See the comment at the top of server.py for details.
Setup
git clone <this-repo>
cd remotive-mcp-server
python3 -m venv .venv
.venv/bin/pip install -r requirements.txt
Run standalone (stdio)
.venv/bin/python server.py
Connect to Claude Desktop / Claude Code
Add to your MCP client config (e.g. claude_desktop_config.json):
{
"mcpServers": {
"remotive": {
"command": "/absolute/path/to/remotive-mcp-server/.venv/bin/python",
"args": ["/absolute/path/to/remotive-mcp-server/server.py"]
}
}
}
Restart your client and the remotive_search_jobs / remotive_get_job tools will be available.
Evaluation
evaluation/evaluation.xml has 10 verifiable question/answer pairs exercising both tools (including ones that require calling remotive_search_jobs to find a job id, then remotive_get_job to read facts only present in the full description). Remotive's listing changes daily, so the pinned answers are a snapshot from 2026-08-10 — regenerate them against the live API before relying on this for regression testing.
Attribution & fair use
Per Remotive's API terms, every tool response includes a link back to the original Remotive job URL and credits Remotive as the source. This server does not cache or redistribute job data beyond a single request/response cycle, and calls the endpoint only on demand (Remotive asks for at most a few requests per day since data refreshes slowly).
Stack
Python 3.10+, MCP Python SDK, httpx, pydantic.
License
MIT — see LICENSE.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。