Ambee MCP Server is Ambee's official Model Context Protocol (MCP) server, used to let any MCP-compatible AI assistant (such as Claude, ChatGPT, Cursor, VS Code, Ollama, etc.) directly access real-time air quality, pollen, and weather data. After users point their assistant to this server, the assistant can answer questions such as "Is tree pollen severe in Atlanta today?" or "Should Saturday's outdoor activity be moved indoors?" by directly calling Ambee data, instead of relying on guesses from outdated training data.
Purpose Connect Ambee's real-time environmental data to MCP-compatible AI assistants, enabling AI to answer questions about air quality, pollen, and weather based on real data.
Main capabilities Provides six tools covering the latest conditions and 48-hour forecasts for three types of datasets (air quality, pollen, weather). Each tool accepts one of the following location parameters: lat + lng (numbers), or place (free text, for example "Bengaluru"). Sending both or neither is rejected by the client before the request. All /v3 endpoints natively support place, so no separate geocoding step is needed. Optional parameters include: locale (boolean) — adds a localTime field to each record; aqiStandard (air quality only) — EPA (default), IN, UK, CN, or CA; units (weather only) — imperial (default), metric, or si; speciesRisk (pollen only) — includes species-level risk levels when supported by the region. Every response is checked against the status codes in the Ambee documentation. All non-2xx responses are returned as MCP tool errors (isError: true), with a readable message combining Ambee's own error text and a brief follow-up action hint, so the calling assistant can explain the reason for the failure rather than merely displaying the raw status code.
Usage Requires an Ambee API Key, which can be obtained for free from the Ambee API dashboard. Never commit keys or put them in configuration files under version control; see SECURITY.md for details. The server runs over stdio, and after startup you should see [ambee-mcp-server] running on stdio on stderr; the processwill wait for MCP messages. The server uses standard MCP over stdio, so it can work with any compatible client. The command and environment variables are the same everywhere; only the configuration file format differs.
Key configuration points for each client:
- Claude Desktop: add to claude_desktop_config.json.
- Cursor: add to ~/.cursor/mcp.json (or project-level .cursor/mcp.json).
- VS Code (GitHub Copilot / MCP extension): add to .vscode/mcp.json.
- Windsurf: add to ~/.codeium/windsurf/mcp_config.json.
- Cline (VS Code extension): add a new server in Cline's MCP settings panel, with Args as /absolute/path/to/ambee-mcp-server/src/index.js and the environment variable as AMBEE_API_KEY=your-ambee-api-key; or directly edit cline_mcp_settings.json using the same command / args / env structure as the other examples.
- Zed: add under context_servers in Zed's settings.json.
- Ollama: The native Ollama application does not yet directly support MCP, but any Ollama model can call this server through an MCP-aware bridge or proxy framework, because Ollama exposes an OpenAI-compatible API, and MCP client libraries (such as mcp-use or LangChain's MCP adapter) can route tool calls through it. The documentation provides a minimal example using mcp-use with a local Ollama model.
- ChatGPT (developer mode / custom connector): ChatGPT's MCP connector support (developer mode, Team/Enterprise workspaces) expects a remote HTTP server rather than a local stdio process. To expose this server in this way, run it behind an MCP-to-HTTP gateway (such as mcp-remote or a small reverse proxy), and register the generated URL as a custom connector in ChatGPT settings.
- Any other MCP client: every MCP client ultimately needs the same three things—the command, args, and env blocks; consult the client documentation for where exactly to place the configuration block.
Example prompts "How bad is the air quality in Los Angeles right now?" "When will PM2.5 be lowest tomorrow so I can go for a run?" "My child is allergic to grass—will Saturday morning be bad for her in Austin?" "What is the current UV index in 90210, in metric units?" "Compare the air quality at our Bangalore and Austin offices."
Use cases Suitable for AI assistant use cases that need to answer questions based on real-time air quality, pollen, and weather data, such as outdoor activity planning, allergy risk queries, air quality comparisons, UV index queries, and more.
Limitations All tools are read-only; they cannot modify the user's Ambee account or data. MCP calls share the same Ambee API quota/rate limits as REST calls. Currently only "latest" and "48-hour forecast" are exposed. Ambee's /v3 API also supports historical data and a 120-hour forecast—check the open issue or submit a feature request.
Support and contributions For questions/bugs, please submit to GitHub Issues; for API support, see support.getambee.com; for API status, see docs.ambeedata.com/api-status; for security reports, see SECURITY.md. Contributions are welcome; see CONTRIBUTING.md for details.
