TensorLink-AI/ephemeris-mcp

Overview

MCP server for time-series foundation model (TSFM) forecasting. Provides probabilistic forecasts from Chronos-2, TimesFM 2.5, Toto 2, TiRex-2, and IBM Granite models, or an accuracy-weighted ensemble. Built in JavaScript.

Best for

Best for
Developers needing an MCP-based bridge to multiple time-series forecasting models.

Use cases

• Generate probabilistic forecasts from multiple TSFM models via MCP
• Combine model outputs into an accuracy-weighted ensemble
• Integrate time-series forecasting into MCP-compatible clients

How to use

Install

npx -y ephemeris-mcp

Tools exposed

• forecast
• list_models
• get_balance
• get_usage

Tested with

Claude Desktop, Claude Code, Cursor, VS Code, ChatGPT

Example client config

{\n  "mcpServers": {\n    "ephemeris": {\n      "url": "https://ephemeris.cascade.industries/api/mcp",\n      "headers": { "Authorization": "Bearer pc_live_your_key" }\n    }\n  }\n}

Notes

MCP server for time-series foundation model (TSFM) forecasting. Provides probabilistic forecasts from Chronos-2, TimesFM 2.5, Toto 2, TiRex-2, and IBM Granite models, or an accuracy-weighted ensemble. Built in JavaScript.

0 stars on GitHub. Last updated 2026-10-07. Licensed MIT.

Use cases

• Generate probabilistic forecasts from multiple TSFM models via MCP
• Combine model outputs into an accuracy-weighted ensemble
• Integrate time-series forecasting into MCP-compatible clients

Pros

• Supports several leading time-series foundation models
• Offers ensemble forecasting for potentially better accuracy
• Probabilistic outputs for uncertainty-aware predictions

Cons

• Zero GitHub stars suggests early-stage or unproven project
• No single vendor backing (vendor listed as Various)
• JavaScript ecosystem may limit access to Python-native TSFM tooling

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.