Summary
Research stops running inside your agent's tool loop.
The entire surface is one tool with one parameter, so no part of the retrieval strategy has to live in your prompt or your token budget. What comes back is a finished report, which makes this a substitute for a research subagent rather than another search tool to feed one.
What it is
A single research agent behind a single tool. It takes a prompt, aggregates multiple sources, scrapes the web and synthesises what it finds, returning a report rather than search results.
What you get
- One parameter — the prompt — with source selection, scraping and synthesis handled on Octagon's side
- A written report as the return value, rather than links for your agent to fetch and read
- The research shapes the agent is built around: competitive analysis, market intelligence, academic synthesis, technical analysis, policy research and trend analysis
Requirements
An Octagon account with its API key in OCTAGON_API_KEY, and npx on your PATH.
Setup effort
One command plus a key — env OCTAGON_API_KEY=your_octagon_api_key npx -y octagon-deep-research-mcp, then supply credentials