Repolode
Submit
jordan-gibbs

jordan-gibbs/hyperresearch

A deep research skill and pipeline for Claude Code and OpenAI Codex that gathers web and scholarly sources into a persistent, searchable markdown vault and writes cited reports.

Open-source alternative to:Perplexity

View on GitHubWebsite
Stars
3,884
Forks
408
Open issues
35
Last commit
9 days ago

What is it?

Hyperresearch adds a multi-step deep research pipeline to Claude Code and OpenAI Codex. A single prompt is decomposed into sub-questions, searched across the web and scholarly sources such as OpenAlex, Crossref, CORE, SEC EDGAR and FRED, then drafted, critiqued and cite-checked before a final report is written with source provenance.

Every fetched source is stored as a markdown note with YAML frontmatter in a SQLite-indexed vault. The markdown files are the source of truth and the index can be rebuilt from them, so later sessions search existing material before fetching new pages. Runs are resumable, can be capped by a spend budget, and pass verification checks such as quote integrity and retraction checks before shipping.

The project is installed with pip and set up per project with a single command, with optional MCP server and local web UI for browsing the vault from other tools.

Key features

  • 01Runs a multi-step research pipeline from one prompt, with critics and a cite-check before the report ships.
  • 02Searches web pages plus scholarly and data sources including OpenAlex, Crossref, CORE, SEC EDGAR and FRED.
  • 03Stores every fetched source as searchable markdown in a SQLite-indexed vault that later sessions reuse.
  • 04Resumes crashed runs at the exact step where they stopped and supports spend budgets.
  • 05Attempts to recover legal open-access full text for paywalled papers via Unpaywall, Europe PMC and CORE.
  • 06Exposes the vault through an MCP server and a local web UI.

Use cases

  • Producing a cited literature review or competitor analysis from web and filings in one session.
  • Building a persistent research wiki that grows across multiple projects.
  • Letting Claude Desktop or Cursor query a markdown research vault through MCP.
  • Scholarly literature searches that include books and non-STEM fields.

Quick start

  1. 1

    Install the Python package

    pip install hyperresearch
  2. 2

    Install the pipeline into your project for Claude Code

    hyperresearch install
  3. 3

    In Claude Code, run the skill with your question

    /hyperresearch <anything>

Things to know

  • Codex runs need explicit sandbox and network flags, and Codex has no browser lane, so login- or bot-blocked pages are left for you to review.
  • Codex tool locks are only instructions, not enforced restrictions.
  • The dissertation tier is experimental and its length and citation targets are not yet wired in.

Topics

#deep-research#claude-code#ai-agent#research#citations#mcp#knowledge-base#rag#agents#agentskills

FAQ

What is hyperresearch?
A deep research skill and pipeline for Claude Code and OpenAI Codex that gathers web and scholarly sources into a persistent, searchable markdown vault and writes cited reports. Hyperresearch adds a multi-step deep research pipeline to Claude Code and OpenAI Codex. A single prompt is decomposed into sub-questions, searched across the web and scholarly sources such as OpenAlex, Crossref, CORE, SEC EDGAR and FRED, then drafted, critiqued and cite-checked before a final report is written with source provenance.
Is hyperresearch free to use? What license does it have?
hyperresearch is open source under the MIT license. Check the license terms before using it in commercial products.
Is hyperresearch still maintained?
Actively maintained. The last commit to hyperresearch was on Oct 10, 2026.
What can I use hyperresearch for?
Producing a cited literature review or competitor analysis from web and filings in one session. Building a persistent research wiki that grows across multiple projects. Letting Claude Desktop or Cursor query a markdown research vault through MCP. Scholarly literature searches that include books and non-STEM fields.
How do I install hyperresearch?
Install the Python package: pip install hyperresearch Install the pipeline into your project for Claude Code: hyperresearch install In Claude Code, run the skill with your question: /hyperresearch <anything>
Where can I get hyperresearch?
The source code and installation instructions are on GitHub: https://github.com/jordan-gibbs/hyperresearch

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects