# skforecast-ai MCP server

> Forecast time series in local CSV files with deterministic skforecast workflows and the script that produced each result.

- Listing: https://mcp.tc/i/skforecast-ai
- Connect: this is a local (stdio) server; install it on your machine (see Install). The listing link is a page, not an MCP endpoint.
- Type: local (stdio)
- Auth: no API key needed
- Category: [Data & Analytics](https://mcp.tc/c/analytics)
- Vendor: skforecast
- Homepage: <https://ai.skforecast.org/stable/user-guides/mcp-server.html>
- Docs: <https://ai.skforecast.org>
- Repository: <https://github.com/skforecast/skforecast-ai>
- Package: pypi `skforecast-ai-mcp`

## About

Gives coding agents tools for time series forecasting with skforecast. The agent can profile a CSV file, plan a forecaster, backtest it, compare candidate models and produce forecasts. Decisions come from rule-based code, each result comes with the script that produced it, and responses do not carry rows of your data.

It runs locally over stdio, started with uvx from the skforecast-ai-mcp PyPI package. It needs one directory passed with --allow-dir and only reads CSV files inside it. No API key or sign-in is needed; the language model comes from the MCP client.

## What it can do

- Profile a CSV file containing time series data
- Plan a forecaster for a dataset
- Backtest a forecaster
- Compare candidate models
- Generate forecasts
- Return the reproducible script behind each result
- Restrict file access to one allowed directory

## Example prompts

- "Profile sales.csv and tell me what kind of forecaster fits it."
- "Backtest a forecaster on demand.csv and show the script that ran."
- "Compare candidate models for my hourly energy data and pick the best."
- "Forecast the next 30 days from daily\_visits.csv."

## Install

### Claude Code

1. Run this in a terminal, in your project folder:

```bash
claude mcp add --transport stdio skforecast-ai -- uvx skforecast-ai-mcp --allow-dir /path/to/allow-dir
```

2. Start Claude Code and type `/mcp`. **skforecast-ai** should show as connected.

Add `--scope user` to make it available in every project.

### Claude Desktop

1. Open **Settings → Developer → Edit Config**. It opens `claude_desktop_config.json`. Add:

`claude_desktop_config.json`:

```json
{
  "mcpServers": {
    "skforecast-ai": {
      "command": "uvx",
      "args": [
        "skforecast-ai-mcp",
        "--allow-dir",
        "/path/to/allow-dir"
      ]
    }
  }
}
```

2. Save the file and restart Claude Desktop.

Needs uv (Python) on your computer. The file lives in `~/Library/Application Support/Claude/` on macOS and `%APPDATA%\Claude\` on Windows.

### Cursor

[Add to Cursor](<https://cursor.com/install-mcp?name=skforecast-ai&config=eyJjb21tYW5kIjoidXZ4IiwiYXJncyI6WyJza2ZvcmVjYXN0LWFpLW1jcCIsIi0tYWxsb3ctZGlyIiwiL3BhdGgvdG8vYWxsb3ctZGlyIl19>) (opens Cursor)

Or add it by hand to `~/.cursor/mcp.json` (all projects) or `.cursor/mcp.json` (this project):

`mcp.json`:

```json
{
  "mcpServers": {
    "skforecast-ai": {
      "command": "uvx",
      "args": [
        "skforecast-ai-mcp",
        "--allow-dir",
        "/path/to/allow-dir"
      ]
    }
  }
}
```

Needs uv (Python) on your computer.

### VS Code

[Install in VS Code](<https://vscode.dev/redirect/mcp/install?name=skforecast-ai&config=%7B%22type%22%3A%22stdio%22%2C%22command%22%3A%22uvx%22%2C%22args%22%3A%5B%22skforecast-ai-mcp%22%2C%22--allow-dir%22%2C%22%2Fpath%2Fto%2Fallow-dir%22%5D%7D>) (opens VS Code)

Or from a terminal:

```bash
code --add-mcp '{"name":"skforecast-ai","type":"stdio","command":"uvx","args":["skforecast-ai-mcp","--allow-dir","/path/to/allow-dir"]}'
```

Or commit it to the repo in `.vscode/mcp.json`:

`.vscode/mcp.json`:

```json
{
  "servers": {
    "skforecast-ai": {
      "type": "stdio",
      "command": "uvx",
      "args": [
        "skforecast-ai-mcp",
        "--allow-dir",
        "/path/to/allow-dir"
      ]
    }
  }
}
```

Needs uv (Python) on your computer.

### Devin Desktop

1. Add it to `~/.config/devin/mcp_config.json` (macOS and Linux) or `%APPDATA%\devin\mcp_config.json` (Windows):

`mcp_config.json`:

```json
{
  "mcpServers": {
    "skforecast-ai": {
      "command": "uvx",
      "args": [
        "skforecast-ai-mcp",
        "--allow-dir",
        "/path/to/allow-dir"
      ]
    }
  }
}
```

2. Refresh the MCP server list in Cascade.

Devin Desktop is the new name for Windsurf.

### Codex

```bash
codex mcp add skforecast-ai -- uvx skforecast-ai-mcp --allow-dir /path/to/allow-dir
```

Or edit `~/.codex/config.toml` directly:

`config.toml`:

```toml
[mcp_servers.skforecast-ai]
command = "uvx"
args = ["skforecast-ai-mcp", "--allow-dir", "/path/to/allow-dir"]
```

Needs uv (Python) on your computer.

### Gemini CLI

```bash
gemini mcp add skforecast-ai uvx -- skforecast-ai-mcp --allow-dir /path/to/allow-dir
```

This adds it to the current project. Add `-s user` to use it everywhere.

### Any client

Most clients that start local servers accept this shape:

```json
{
  "mcpServers": {
    "skforecast-ai": {
      "command": "uvx",
      "args": [
        "skforecast-ai-mcp",
        "--allow-dir",
        "/path/to/allow-dir"
      ]
    }
  }
}
```

Zed puts servers under `context_servers` in its settings, with the same `command`, `args` and `env` fields.

Needs uv (Python) on your computer.

## Recent changes

- 2026-11-11 20:33 UTC: New version 0.5.0 (was 0.4.2), read from PyPI. Source: automatic check.
- 2026-10-18 04:21 UTC: Tracking started: version 0.4.2. Source: automatic check.

More on the History tab, with descriptions before and after for the newest entries: https://mcp.tc/i/skforecast-ai#history (Atom feed: https://mcp.tc/i/skforecast-ai/history.xml). It records what mcp.tc's anonymous checks saw, every 3 hours where the tool list can be read without sign-in; it is not a security review.

## Details

- Server version: 0.5.0
- Last checked: 2026-10-18
- Listed: 2026-10-18
- Updated: 2026-10-26

---
Source: https://mcp.tc/i/skforecast-ai (mcp.tc is an independent directory, not affiliated with this server's publisher). Corrections: https://mcp.tc/report
