MCP (Model Context Protocol) lets AI agents connect directly to your cap table. Learn what MCP is, how it works, and why AI-native cap table management is replacing manual workflows.
MCP (Model Context Protocol) is an open standard that lets AI agents connect to external tools, databases, and APIs through a universal interface. For cap table management, MCP means your AI assistant can directly query shareholders, model dilution scenarios, generate documents, and update equity records — all through natural language commands instead of manual clicks.
This guide explains what MCP is, how it works, and why it is becoming essential infrastructure for modern cap table management.
What Is MCP?
Model Context Protocol is an open specification originally developed by Anthropic that standardizes how AI models interact with external data sources and tools. Think of it as a USB-C port for AI — a single, universal connector that lets any AI agent talk to any compatible service.
Before MCP, every AI integration required custom code. If you wanted Claude to read your cap table, you needed a bespoke integration. If you then wanted GPT-4 to do the same thing, you needed a completely different integration. MCP eliminates this fragmentation by providing one protocol that works across all AI clients.
How MCP Works
The protocol follows a client-server architecture:
| Component | Role | Example |
|---|---|---|
| MCP Client | The AI application requesting data or actions | Claude, Cursor, your custom AI agent |
| MCP Server | The service exposing tools and data | OpenCap Stack MCP server |
| Transport | The communication layer | stdio, HTTP, SSE |
When an AI agent needs cap table data, the conversation looks like this:
- The user asks their AI assistant a question about equity
- The AI client discovers available MCP tools from connected servers
- The client calls the appropriate tool (e.g.,
cap_table_summary) - The MCP server executes the request against the actual database
- Results flow back to the AI, which formats the answer for the user
The key insight is that the AI never needs to be trained on your specific data. It accesses live, real-time information through structured tool calls.
Why Cap Tables Need MCP
Cap table management is one of the highest-value use cases for MCP because equity data is complex, constantly changing, and frequently queried by people who are not equity specialists.
The Problem With Traditional Cap Table Software
Traditional cap table tools require users to log in, navigate menus, find the right report, and interpret the output. This creates friction in three common scenarios:
Founder answering investor questions. During fundraising, founders field dozens of questions about ownership percentages, dilution, vesting schedules, and SAFE conversion terms. Each answer requires logging into the cap table tool, running a report, and relaying the numbers. With MCP, the founder asks their AI assistant directly: "What would ownership look like after a $3M Series A at a $15M pre-money valuation?"
Lawyer reviewing equity agreements. Legal teams need to cross-reference cap table data against corporate documents during due diligence. Instead of requesting exports and manually checking numbers, an MCP-connected AI agent can pull live cap table data while reviewing agreements.
CFO preparing board materials. Finance teams regularly need equity summaries, option pool status, and dilution projections for board presentations. MCP lets them generate these reports through conversational queries rather than manual data extraction.
What MCP Enables for Cap Tables
With an MCP-connected cap table, AI agents can:
- •Query ownership in real time — "Who are our top 10 shareholders by percentage?"
- •Model fundraising scenarios — "Show me dilution if we raise $5M at a $20M cap"
- •Track vesting schedules — "How many options vest for the engineering team this quarter?"
- •Generate documents — "Draft a board consent for a 1,000-share option grant to our new VP of Sales"
- •Monitor compliance — "Is our 409A valuation still current?"
- •Search and retrieve documents — "Find the SAFE agreement for Acme Ventures"
- •Calculate waterfall distributions — "What does each shareholder receive at a $50M exit?"
These are not hypothetical features. They are live MCP tools available today in OpenCap Stack.
MCP vs Traditional API Integrations
You might wonder how MCP differs from a standard REST API. Both let external systems access your cap table data, but they serve fundamentally different purposes.
| Aspect | REST API | MCP |
|---|---|---|
| Consumer | Developers writing code | AI agents acting on user instructions |
| Discovery | Read API documentation | Automatic tool discovery at runtime |
| Interface | HTTP endpoints with JSON | Structured tool calls with schemas |
| Flexibility | Fixed endpoints | Dynamic tool composition |
| User interaction | None — code runs in the background | Conversational — user sees results in natural language |
| Setup required | Write integration code | Connect MCP server and start asking questions |
REST APIs are still essential for programmatic integrations — syncing data between systems, building dashboards, or automating workflows. MCP is the layer that makes your cap table accessible to AI agents and their human operators.
The two approaches complement each other. OpenCap Stack provides both a full REST API and an MCP server, so you can build traditional integrations and use AI agents against the same data.
How OpenCap Stack's MCP Server Works
OpenCap Stack ships an MCP server with 29 tools that cover the full lifecycle of cap table management. Here is how the major tool categories break down:
Equity and Ownership
- •
cap_table_summary— full ownership breakdown across all share classes - •
get_fully_diluted_shares— total share count including all convertible instruments - •
list_stakeholders/get_stakeholder— query and inspect individual equity holders - •
list_share_classes/get_share_class— view share class details and rights - •
calculate_dilution— model dilution for a proposed round
Fundraising and Instruments
- •
list_safes/get_safe/create_safe— manage SAFE agreements - •
run_waterfall_analysis— distribution modeling at various exit prices - •
get_latest_valuation/get_valuation_history— 409A and fair market value tracking - •
cross_company_dilution— compare dilution across portfolio companies
Documents and Compliance
- •
list_documents/search_documents— find and retrieve corporate documents - •
generate_document— create new equity-related documents - •
list_data_rooms/create_data_room— manage investor data rooms - •
get_83b_status/send_83b_reminder— track 83(b) election compliance
Equity Plans and Grants
- •
list_equity_plans/create_equity_plan— manage option pools and equity incentive plans - •
list_equity_grants/create_equity_grant— issue and track individual grants - •
get_vesting_schedule— view vesting status for any grant
Connecting to the MCP Server
To connect any MCP-compatible AI client to your OpenCap Stack instance, you add the server configuration:
{
"mcpServers": {
"opencap": {
"url": "https://api.opencapstack.com/mcp",
"headers": {
"Authorization": "Bearer YOUR_API_KEY"
}
}
}
}Once connected, your AI assistant automatically discovers all available tools and can start answering equity questions using live data.
Real-World MCP Use Cases
Due Diligence Preparation
When preparing for investor due diligence, an AI agent with MCP access can:
- Pull the current fully diluted cap table
- List all outstanding SAFEs and convertible notes
- Check that the 409A valuation is current
- Search for missing documents (unsigned agreements, missing 83(b) elections)
- Generate a summary report for the data room
What previously took hours of manual work becomes a five-minute conversation.
Board Meeting Prep
Before a board meeting, a founder can ask their AI assistant to generate an equity summary including option pool utilization, recent grants, vesting milestones, and dilution projections — all sourced directly from the live cap table through MCP.
Employee Equity Questions
HR teams and managers frequently field questions from employees about their equity. An MCP-connected assistant can instantly answer questions like "How many of my options have vested?" or "What is my current strike price?" by querying the actual grant records.
The Future of AI-Native Cap Table Management
MCP represents a fundamental shift in how people interact with financial data. Instead of learning complex software interfaces, users describe what they need in plain language and AI agents handle the data retrieval, computation, and formatting.
For cap table management specifically, this shift matters because:
- Equity is cross-functional. Founders, lawyers, accountants, HR teams, and investors all need cap table data, but each group uses different tools. MCP lets every stakeholder access cap table data through whatever AI assistant they already use.
- Queries are often ad hoc. Unlike accounting systems where reports are standardized, cap table questions vary wildly. "What happens to employee options if we get acquired at $30M?" is not a standard report — but it is a natural MCP query.
- Accuracy is critical. Cap table errors have legal and financial consequences. MCP tools query the authoritative database directly, eliminating the copy-paste errors that plague spreadsheet-based workflows.
How OpenCap Stack Helps
OpenCap Stack is built from the ground up as an AI-native cap table platform. The MCP server is not a bolt-on feature — it is a core part of the product architecture.
- •29 MCP tools covering equity, fundraising, documents, compliance, and reporting
- •Available on all plans — MCP access is not locked behind an enterprise paywall
- •Open source — the MCP server code is MIT-licensed, so you can inspect, modify, and self-host
- •Full REST API — 60+ endpoints for traditional programmatic integrations alongside MCP
- •Real-time data — MCP tools query live data, not cached exports
Connect your AI agent to OpenCap Stack →
FAQ
What AI clients work with OpenCap Stack's MCP server?
Any MCP-compatible client works, including Claude (via Claude Code or the Claude desktop app), Cursor, Windsurf, and custom AI agents built with the Anthropic SDK or OpenAI SDK with MCP support. The protocol is model-agnostic — it works with any LLM.
Is MCP secure for sensitive equity data?
Yes. MCP connections are authenticated with API keys, and all communication happens over HTTPS. The MCP server enforces the same role-based access controls as the web interface — a user's AI agent can only access data that user is authorized to see.
Do I need to be a developer to use MCP with my cap table?
No. If you use an MCP-compatible AI assistant like Claude, you just add the server configuration and start asking questions. There is no code to write. Developers can also use the MCP server programmatically for more advanced integrations.
How is MCP different from just asking ChatGPT about my cap table?
General-purpose AI models do not have access to your actual cap table data. They can only give generic advice based on training data. MCP connects the AI to your real equity records, so answers are based on your actual shareholders, share counts, vesting schedules, and valuations — not hypotheticals.
Is MCP an Anthropic-only technology?
No. While Anthropic developed the initial specification, MCP is an open standard. Any AI provider or tool builder can implement MCP clients and servers. The protocol is designed to be model-agnostic and vendor-neutral.
What happens to my data when I use MCP?
Your data stays in your OpenCap Stack database. MCP tools read from and write to your existing data store — they do not copy your data to a third-party AI service. The AI model receives only the specific query results it requests, not your entire database.
