Startup Valuation MCP Server

Startup Valuation Engine

A comprehensive startup valuation library implementing 80+ formulas from the Startup Valuation textbook — Python library, MCP server, and AI-agent skills.

[CI] [MCP Registry] [Glama] [MCP tools] [PyPI] [Python 3.10+] [Coverage] [Docs] [OpenSSF Scorecard] [License: MIT]

Overview

A production-grade Python library for startup valuation, implementing every formula from the Startup Valuation textbook by Simon Mak (Valuation in Practice Series, Ascent Partners). Designed for developers, financial analysts, and AI agents who need auditable, structured valuation computations.

Three-layer architecture:

graph TB    subgraph Library["Python Library"]        MOD["14 Modules<br/>80+ Functions"] --> VR["ValuationResult"]    end    subgraph MCP["MCP Server"]        VR --> SVR["FastMCP Server<br/>14 Tools"]    end    subgraph Skills["AI-agent skills"]        SVR --> CORE["Core"]        SVR --> ADV["Advanced"]        SVR --> IND["Industry"]        SVR --> STAKE["Stakeholder"]        SVR --> EMER["Emerging"]    end    style Library fill:#0083AB,color:#fff    style MCP fill:#4CAF50,color:#fff    style Skills fill:#9C27B0,color:#fff

1. Python Library — 14 modules, 80+ typed functions, all returning ValuationResult (value + assumptions + sensitivity)
2. MCP Server — 14 folded tools (80+ formulas) for AI agents via stdio and hosted Streamable HTTP
3. AI-agent skills — 6 skill definitions with workflow guidance for valuation domains

Installation

pip install startup-valuation          # library onlypip install startup-valuation[mcp]     # + MCP serverpip install startup-valuation[dev]     # + pytest, ruff, mypy

Quick Start (≤ 5 minutes)

Fastest path: one line — pip install "startup-valuation[mcp]" — then call any *_valuation tool (Scorecard, VC method, Black-Scholes, scenario analysis).

Python Library

from startup_valuation.core import scorecard_valuation, vc_method_post_moneyfrom startup_valuation.advanced import black_scholes, scenario_analysisfrom startup_valuation.types import Scenario
# Scorecard Method (pre-revenue startups)result = scorecard_valuation(    average_valuation=1_500_000,    weights=[0.30, 0.25, 0.15, 0.10, 0.10, 0.05, 0.05],    scores=[1.25, 1.50, 1.20, 0.75, 1.00, 0.90, 1.00],)print(f"Scorecard: ${result.value:,.0f}")  # $1,800,000
# Black-Scholes for real options (startup equity)result = black_scholes(    underlying=20_000_000, strike=5_000_000,    risk_free_rate=0.05, volatility=0.40, time_to_maturity=1.0,)print(f"Option value: ${result.value:,.0f}")  # $15,240,000
# Scenario Analysisscenarios = [    Scenario("bull", 0.20, 10_000_000),    Scenario("base", 0.60, 5_000_000),    Scenario("bear", 0.20, 1_000_000),]result = scenario_analysis(scenarios)print(f"Expected value: ${result.value:,.0f}")  # $5,200,000

MCP Server (for AI Agents)

The server exposes 14 tools, each folding a family of formulas behind a method argument — probability, time value, CAPM, core pre-revenue methods, options, comparables, SaaS, marketplaces, fintech, biotech, hardware, international, stakeholder equity, emerging methods, and a triangulated full analysis.

Local (stdio):

pip install "startup-valuation[mcp]"startup-valuation-mcp          # console script installed with the [mcp] extra
**Prompts and resources.** Besides the 14 tools, the server offers three guidedprompts (`value_pre_revenue_startup`, `value_saas_startup`, `model_funding_round`)and a machine-readable method catalog at `startup-valuation://methods`, so agentscan see every method's required parameters before calling a tool.
# or: python -m startup_valuation.mcp# or ephemeral, no clone: uvx --from startup-valuation startup-valuation-mcp

Hosted (Streamable HTTP) — no install, no API key:

https://startup-valuation.ascent-partners.com/api

OpenCode — add to opencode.json:

"startup-valuation": {  "type": "remote",  "url": "https://startup-valuation.ascent-partners.com/api",  "timeout": 60000}

Claude Desktop / Cursor — add the HTTP URL https://startup-valuation.ascent-partners.com/api as an MCP server, or run the stdio entrypoint above.

MCP Registry — published as com.ascent-partners/startup-valuation (manifest: server.json) and listed on Glama and the Official MCP Registry. The glama.json file holds the Glama maintainer entry.

AI-agent skills

Copy the skills/ directory to your agent's skills folder:

• valuation-core — Scorecard, Berkus, VC Method, Risk Factor Summation
• valuation-foundations — Probability, time value, CAPM, comparables
• valuation-advanced — Black-Scholes, Binomial, Monte Carlo, Scenario Analysis
• valuation-industry — SaaS, Biotech, Fintech, Marketplace, Hardware
• valuation-stakeholder — Dilution, OPM, PWERM, Liquidation Preference
• valuation-emerging — SAFE, Crypto (MV=PQ), ESG, Metcalfe's Law

Valuation Methods by Category

The valuation ladder

flowchart LR    IDEA["Idea / pre-revenue"] --> SC["Scorecard · Berkus · Risk Factor"]    SC --> VC["VC Method · OPM · PWERM"]    VC --> ADV["Black-Scholes · Binomial · Monte Carlo · Scenario"]    ADV --> COMP["Comparables: P/E · P/S · EV/EBITDA"]    COMP --> IND["Industry: SaaS · Biotech · Fintech · Marketplace · Hardware"]    IND --> R["ValuationResult (value · assumptions · sensitivity)"]

Distribution

flowchart LR    PKG["startup-valuation<br/>14 modules · 80+ functions"] --> L["pip install startup-valuation"]    PKG --> MP["pip install startup-valuation[mcp]<br/>→ 14-tool MCP server"]    MP --> T1["stdio (local)"]    MP --> T2["hosted Streamable HTTP"]

Why This Library?

• Auditable — Every function returns ValuationResult with value, method, inputs, assumptions, and sensitivity analysis
• Textbook-accurate — All formulas verified against book example values with unit tests
• AI-ready — MCP server and Skills for seamless AI agent integration
• Industry-specific — Dedicated modules for SaaS, biotech, fintech, marketplace, and hardware startups
• Open source — MIT license, extensible, well-documented

Development

# Install dev dependenciespip install -e ".[dev]"
# Run testspytest
# Run with coveragepytest --cov=startup_valuation --cov-report=term-missing
# Lintruff check .
# Type checkmypy src/startup_valuation --ignore-missing-imports

Note: CI and deployment run through the APG workbench pipeline.

Documentation

• API Reference: GitHub Pages
• Wiki (Theory & Derivations): GitHub Wiki
• PyPI: pypi.org/project/startup-valuation
• Chapter Index: Maps every function to its textbook chapter
• Examples: Interactive code snippets for each valuation category

Companion Textbook

Startup Valuation: A Comprehensive Guide to Valuing Fast-Growing Pre-Revenue Companies
Theory, Methods, Regulation, and Practice — Valuation in Practice Series by Ascent Partners
By Simon Mak · 338 pages · 15 chapters · 300+ exercises · 20+ real-world cases

Citing This Project

@software{startup_valuation_engine,  author = {Mak, Simon},  title = {Startup Valuation Engine},  year = {2026},  url = {https://github.com/simonmak-ascent/startup-valuation},  license = {MIT},}

Based on formulas from the Startup Valuation textbook.

Use with Context7

Up-to-date Startup Valuation Engine documentation is indexed on Context7, so coding agents can pull it into context on demand. With the Context7 MCP server or ctx7 CLI installed, name the library in your prompt:

use library /simonmak-ascent/startup-valuation for API and docs

License

MIT — see LICENSE.

By Ascent Partners — part of the Valuation in Practice Series.

If this saves you time, a ⭐ on GitHub helps others find it.