如何将 Exa 远程 MCP 连接到 VS Code

约 1 分钟即可把 Exa 的远程 MCP 服务端接入 VS Code。如需排查 OAuth / 401 问题,可先用 OAuth 2.1 调试器体检端点。

🛡️ Official✓ VerifiedStreamable HTTPAPI Key🛡️ A+~361 tok11 tools
Streamable HTTPhttps://mcp.exa.ai/mcp

🔒 Client-side only — never leaves your browser.

Install in VS Code →

Or add to .vscode/mcp.json:

{
  "servers": {
    "exa": {
      "type": "http",
      "url": "https://mcp.exa.ai/mcp",
      "headers": {
        "Authorization": "Bearer YOUR_API_KEY"
      }
    }
  }
}

接入步骤

  1. 打开 MCP 配置 — 使用本页 vscode: deeplink 一键安装,或在项目根目录创建 .vscode/mcp.json。
  2. 添加服务端 — 写入 { "servers": { "exa": { "type": "http", "url": https://mcp.exa.ai/mcp } } }。
  3. 完成认证 — 在 Exa 控制台获取凭据,并将其作为 Authorization: Bearer <密钥> 请求头加入配置。
  4. 验证 — 重启 Copilot Chat 选择 Agent 模式后可看到工具。在 在线 Inspector 中粘贴 https://mcp.exa.ai/mcp 并运行 initialize + tools/list 验证连通性。

🧰 工具预览(11)

ToolSafetyDescriptionParameters
web_search_exa
~32 tok
🟢 ReadSearch the web with Exa neural search.—
get_code_context_exa
~35 tok
🟢 ReadGet code context for programming queries.—
research_paper_search
~30 tok
🟢 ReadSearch academic papers.—
company_research
~28 tok
🟢 ReadResearch companies.—
crawling
~29 tok
🟢 ReadCrawl a URL and extract content.—
competitor_finder
~31 tok
🟢 ReadFind competitors of a company.—
linkedin_search
~33 tok
🟢 ReadSearch LinkedIn profiles and companies.—
wikipedia_search_exa
~29 tok
🟢 ReadSearch Wikipedia.—
github_search
~29 tok
🟢 ReadSearch GitHub repositories.—
deep_researcher_start
~31 tok
🟠 WriteStart a deep research task.—
deep_researcher_check
~34 tok
🟢 ReadCheck status of a deep research task.—

更多连接指南