@calcis/mcp-server

Model Context Protocol server for Calcis.

Live pricing for 25+ models, side-by-side comparisons, and a web estimator: https://calcis.dev

• Full price index: https://calcis.dev/models
• Compare models: https://calcis.dev/compare
• API reference: https://calcis.dev/api-docs

Exposes three tools that AI agents (Claude Desktop, Cursor, Cline, and anything else MCP-compatible) can call directly:

| Tool | What it does | Needs API key? | | --------------- | ------------------------------------------------------------------ | -------------- | | estimate_cost | Pre-flight cost estimate for a prompt + model | yes | | list_models | Returns every supported model with current pricing | no | | compare_models| Side-by-side cost comparison between two models at a token count | no |

list_models and compare_models ship with the pricing dataset embedded, so they work offline.

Install

Once published, the server runs via npx:

npx @calcis/mcp-server

Or install globally:

npm install -g @calcis/mcp-server
calcis-mcp

Get your API key for estimate_cost at calcis.dev/dashboard.

Claude Desktop

Edit ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or the equivalent on other platforms:

{
  "mcpServers": {
    "calcis": {
      "command": "npx",
      "args": ["@calcis/mcp-server"],
      "env": {
        "CALCIS_API_KEY": "calc_your_key_here"
      }
    }
  }
}

Restart Claude Desktop. The Calcis tools will now appear in the available tools list for every conversation.

Cursor

In Settings → Features → Model Context Protocol, add:

• Name: calcis
• Command: npx
• Args: @calcis/mcp-server
• Env: CALCIS_API_KEY=calc_your_key_here

Tools

estimate_cost

Pre-flight estimate for any prompt / model pair.

Input:

{
  "prompt": "Explain quantum computing in three bullets",
  "model": "claude-sonnet-4-6",
  "systemPrompt": "You are a concise technical writer."
}

Returns the Calcis API response JSON (model, inputTokens, outputTokens, inputCost, outputCost, totalCost, confidence).

list_models

No input. Returns { version, generatedAt, source, models[] } where each model has id, displayName, provider, inputPricePerMTok, outputPricePerMTok, contextWindow, maxOutput, and lastUpdated. Browse the same dataset on the web at calcis.dev/models.

compare_models

{
  "modelA": "claude-sonnet-4-6",
  "modelB": "gpt-4o",
  "inputTokens": 1000,
  "outputTokens": 500
}

Returns per-model breakdowns plus deltaTotal (cost(B) − cost(A)) and cheaper (the model ID that wins on total cost). For an interactive, multi-model version see calcis.dev/compare.

Config

| Env var | Required for | Notes | | ----------------- | --------------------------- | -------------------------- | | CALCIS_API_KEY | estimate_cost | Pro plan or above |

The other two tools read the embedded pricing dataset: no API key, no network call.

Links

• Web App
• API Docs
• Pricing dataset (@calcis/pricing)
• CLI (calcis)
• GitHub Action

License

MIT