# npm Trends MCP [Health: Active]

**Category:** 💻 Developer Tools  
**Repository:** https://github.com/trendsmcp-ai/npm-trends-mcp  
**GitHub Stars:** 0  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/npm-trends-mcp-2

## Description
npm package trends and weekly downloads over time. Free key at trendsmcp.ai

## Tools
Capabilities this server exposes over MCP:

- **get_time_series** — Weekly (or daily) history for **one** `source` + `keyword`. Same name on MCP and REST (`mode: "get_time_series"`). REST also accepts `get_trends` as an alias.
- **get_growth** — Point-to-point percent change. Several windows in one call still count as **one** request for that source + keyword.
- **get_top_trends** — Live ranked list. **No keyword.** On MCP, `type` is required and must match the feed name **exactly** (including capitals). On REST, omit `type` only if you intend to pull every feed (billed per feed).

## Claude Desktop Quick Installation
Remote MCP endpoint (confidence: high). Install path detected from listing signals. Add as a URL/SSE server in your client:

```json
"mcpServers": {
  "npm-trends-mcp": {
    "url": "https://api.trendsmcp.ai/mcp"
  }
}
```

## Documentation

## What npm Trends MCP does

The npm Trends MCP MCP server gives MCP-compatible agents access to trend data through the TrendsMCP service. Its three tools cover different query types:

- `get_time_series` returns weekly history, or daily history where supported, for one keyword on one source.
- `get_growth` calculates point-to-point percentage change for one keyword and source. It can compare multiple time windows in a single call.
- `get_top_trends` returns a live ranked feed and does not take a keyword.

The service supports npm-related trend use cases and other listed sources through a shared API. The returned index is normalized from 0 to 100 where the source pipeline supports that format; absolute `volume` values are included when available.

## How it works

The hosted endpoint is `https://api.trendsmcp.ai/mcp`, and requests use the same Bearer API key as the REST endpoint at `https://api.trendsmcp.ai/api`. The repository also includes a stdio adapter for local hosts and Glama. With the local adapter, tool listing works without a key, while paid data requests require `TRENDSMCP_API_KEY` and use the same quota.

Each historical or growth request accepts one source and a keyword. `get_growth` also accepts a comma-separated source list and supports presets from `7D` through `5Y`, along with month-to-date, quarter-to-date, and year-to-date windows. `get_top_trends` requires an exact feed name in MCP requests, including capitalization. It supports pagination, result limits up to 200, categories for some feeds, and ranking or rank-change sorting.

## Setup and configuration

The npm Trends MCP MCP server can be added as a remote HTTP server in clients that support remote MCP connections. Configure the endpoint with an `Authorization` header containing `Bearer YOUR_API_KEY`. The README provides client-specific setup for Cursor, VS Code, Windsurf, Cline, and Claude Code.

Claude Desktop does not use native HTTP in the documented setup. Instead, it launches `mcp-remote` through `npx` and passes the authorization value through an environment variable. The Claude.ai custom connector uses the hosted `www.trendsmcp.ai/mcp` path with OAuth rather than a Bearer key.

For local stdio operation, install the repository package and run `python -m trends_mcp_server`. Add the process to the client configuration and provide `TRENDSMCP_API_KEY` in its environment. A free API key is available from the TrendsMCP account page, with the README stating a 100-request monthly free tier.

## Tools and capabilities

`get_time_series` requires `keyword` and `source`; REST additionally accepts `data_mode` with `weekly` as the default or `daily` where supported. REST recognizes `get_trends` as an alias for this mode.

`get_growth` requires `keyword` and `source`. Its default growth window is `12M`, and custom recent, baseline, and named date objects are also supported.

`get_top_trends` requires `type` on MCP. `limit` defaults to 25, `offset` enables pagination, and `window` applies when sorting by rank change. On REST, leaving out `type` requests every feed and is billed per feed.

## Limitations and notes

Queries are constrained to one source and keyword for time-series requests. Feed names for live rankings must match the documented name exactly on MCP. Data shape varies by source: normalization is not universal, and absolute volume appears only when available. The repository identifies the project under the MIT license.

_Full upstream README: https://allmcps.com/mcp/npm-trends-mcp-2/readme_