Developers

MCP server

The Foundily equity engine as four native tools for any MCP-compatible agent — Claude Desktop, Claude Code, Cursor, or your own agent runtime. No HTTP client code required; the agent calls the tool, the tool calls the API, you get a structured result back in the conversation.

Install

The package runs via npx — nothing to install globally, and your MCP client will keep it up to date automatically.

Terminal

npx @foundily/mcp

Configure

Add Foundily to your MCP client's config — this exact block works for Claude Desktop (claude_desktop_config.json) and Claude Code (.mcp.json, or via claude mcp add). Get an API key from /dashboard/api-keys.

mcp config

{
  "mcpServers": {
    "foundily": {
      "command": "npx",
      "args": ["-y", "@foundily/mcp"],
      "env": {
        "FOUNDILY_API_KEY": "fd_live_..."
      }
    }
  }
}

The key is read from the FOUNDILY_API_KEY environment variable — it never needs to appear in the conversation itself.

Tools

Four tools, one per calculator surface — each a thin wrapper over the public API.

convert_safe

Convert a single SAFE or convertible note at a priced round. Wraps POST /safe/convert.

"Convert a $200k SAFE at a $4M cap into our Series A — $16M pre-money, $4M raised."

model_round

Model a full priced round: founders, an option-pool top-up, and any converting SAFEs. Wraps POST /rounds/model.

"Model a $2M seed at an $8M pre-money with a 10% post-round option pool."

compute_dilution

Run an ordered sequence of SAFEs and priced rounds and trace founder ownership across every stage. Wraps POST /dilution.

"Show my dilution across a pre-seed SAFE and a Series A."

exit_waterfall

Compute exit proceeds per stakeholder given a cap table and an exit value — preferences, seniority, participation and conversion resolved automatically. Wraps POST /waterfall.

"If we sell for $40M, what does each preferred series and common actually get?"

Prefer raw HTTP?

The MCP server is a convenience layer over the same public API — every tool call is a plain HTTP request under the hood. If you're building a custom agent runtime without MCP support, call the REST API directly, or point your model at https://foundily.com/llms.txt for plain-text onboarding.