octagon-vc-agents

Octagon VC Agents provide AI-driven venture capitalist personas that simulate well-known investors. The agents leverage real-time data and research from Octagon Private Markets to enrich their thinking, enabling you to run diligence simulations, pitch feedback, term sheet discussions, and governance-style negotiations in a programmable way. You can interact with individual agents (e.g., octagon-wilson, octagon-hoffman) or run scenarios that combine multiple perspectives to surface biases, valuation considerations, and strategic trade-offs. Use prompts to assign tasks such as deal critique, investor-fit qualification, or metrics reality-check, and the agents will respond with structured analysis and actionable recommendations. To get the most out of them, ensure you provide your OpenAI API key and your Octagon API key in your MCP environment so the agents can access AI capabilities and market data.

Prerequisites:

• Node.js or Python environment not required for this MCP setup since the server is run via pipx in the provided configuration.
• Access to Octagon API and OpenAI API keys.
• pipx installed on your system.

Install steps:

1. Ensure you have pipx installed (example for macOS/Linux):

python3 -m pip install --user pipx
python3 -m pipx ensurepath

2. Obtain your API keys from Octagon and OpenAI and keep them handy.
3. Run the MCP server using pipx with environment keys:

export OPENAI_API_KEY=YOUR_OPENAI_API_KEY_HERE
export OCTAGON_API_KEY=YOUR_OCTAGON_API_KEY_HERE
pipx run --pip-args="--no-cache-dir" octagon-vc-agents run

4. Integrate with your MCP-enabled client (Claude Desktop, Cursor, Windsurf, etc.) by adding the server under mcpServers with the same command and env variables. See the example snippet in the README for your target environment.
5. (Optional) If you are using a specific desktop or tooling integration, configure the corresponding config file to reference the same command and environment variables.

Tips:

• Always provide both OPENAI_API_KEY and OCTAGON_API_KEY in your MCP environment to enable AI reasoning and market data access.
• The command uses pipx to run the octagon-vc-agents package; if you need to run in a restricted environment, ensure network access to PyPI and the Octagon/OpenAI endpoints.
• If you encounter token limits or latency, consider adjusting the model parameters in the underlying agent configuration (not exposed in this MCP snippet) or monitoring API usage quotas.
• When upgrading, re-check the octagon-vc-agents package version and compatibility with your OpenAI model choice.
• For Windows users, the README provides an explicit command line workaround to set environment variables before invoking pipx.