Octagon VC Agents

Octagon VC Agents is an MCP server that runs AI-driven venture capitalist persona agents powered by Octagon Private Markets real-time intelligence. The server connects LLM clients to Octagon Private Markets deal data, private valuations, and research, alongside OpenAI language models. Startup founders, investors, and analysts use it to simulate venture capital thinking and test fundraising materials. It enables users to invoke distinct simulated investor personas—such as Marc Andreessen, Peter Thiel, Reid Hoffman, and Bill Gurley—to perform pitch deck critiques, simulate due diligence, stress-test growth metrics, and prepare for term sheet negotiations. Each investor persona operates based on specialized investment philosophies, decision patterns, and communication styles defined in markdown profiles. Users can evaluate early-stage venture metrics, identify potential portfolio synergies, and gain comparative perspectives on startup funding rounds directly from their AI assistant environment.

Category: Data & Analytics

Tags: investment-analysis, market-intelligence, private-markets, venture-capital

Visit Octagon VC Agents

How to install and configure Octagon VC Agents

1. Obtain an API key from Octagon Private Markets and an API key from OpenAI. 2. Ensure pipx is installed on your operating system (e.g., using brew install pipx on macOS or python -m pip install --user pipx on Windows). 3. Open your client configuration file, such as claude_desktop_config.json for Claude Desktop or model_config.json for Windsurf. 4. Add octagon-vc-agents to the mcpServers section: json { "mcpServers": { "octagon-vc-agents": { "command": "pipx", "args": ["run", "--pip-args=\"--no-cache-dir\"", "octagon-vc-agents", "run"], "env": { "OPENAI_API_KEY": "YOUR_OPENAI_API_KEY_HERE", "OCTAGON_API_KEY": "YOUR_OCTAGON_API_KEY_HERE" } } } } 5. Replace the placeholder values with your active API keys and restart your MCP client.

What you can do with Octagon VC Agents

• Simulate investment committee diligence reviews and identify critical metric gaps that could prevent issuing a term sheet. - Benchmark pre-seed and seed pitch decks against specialized personas to receive pass-or-meet decisions prior to partner meetings. - Generate comparative round evaluations from multiple perspectives, assessing contrarian monopoly advantages against blitzscaling network effects. - Map target customer or design-partner leads across simulated portfolio companies and draft personalized cold outreach emails.

Key facts

• https://github.com/OctagonAI/octagon-vc-agents
• Data & Analytics, Finance, Crypto & Payments
• investment-analysis, market-intelligence, private-markets, venture-capital

Part of MCP Servers

How do I install Octagon VC Agents?

You can run Octagon VC Agents using pipx or install it directly via pip install octagon-vc-agents. To register it in an MCP client like Claude Desktop or Cursor, configure it under your mcpServers settings with the pipx run command, supplying both your OPENAI_API_KEY and OCTAGON_API_KEY in the environment variables block.

What can Octagon VC Agents do?

The server runs AI agents modeled after prominent venture capitalists such as Peter Thiel, Marc Andreessen, and Reid Hoffman. Backed by Octagon Private Markets intelligence, these agents analyze pitch decks, perform simulated due diligence, critique metrics, and provide deal assessments from specific investment philosophies.

Which MCP clients work with Octagon VC Agents?

Octagon VC Agents works with any client supporting the Model Context Protocol over stdio. Documented setups include Claude Desktop, Cursor Desktop version 0.45.6 or later, and Codeium Windsurf.

What API keys are required for Octagon VC Agents?

You require two separate credentials to run the server: an Octagon API key obtained from your Octagon account to access private market intelligence, and an OpenAI API key to power the language model analysis.

Is Octagon VC Agents open source?

Yes, Octagon VC Agents is open source. The project is distributed under the MIT license with source code accessible on GitHub.