Deep Research Agent: automation for Hermes Agent

Hermes skill that runs a seven-phase literature review with DOI checks and an APA 7th Word report

What Deep Research Agent does

The skill formulates a search plan and stops for your approval before doing any work. It then launches three concurrent subagents that query Scopus, Exa, OpenAlex and Semantic Scholar, screens a large pool of abstracts, and downloads and reads the full text of the most relevant papers so claims are not based on abstracts alone. It targets Q1 and Q2 journals and bans Q4 and MDPI publications.

Before writing, it runs live HTTP requests against every generated DOI and cross-checks claims against the raw full text. An included peer-review skill strips filler wording and enforces an academic tone. The output is a Word document with APA 7th hanging indents and generated charts, and the final phase saves summaries to a local Obsidian vault and uploads each cited reference as a separate source to Google NotebookLM. All reports are generated in English regardless of the conversation language.

Key features

Three concurrent subagents search Scopus, Exa, OpenAlex and Semantic Scholar
Two-stage sourcing that reads the full text of highly relevant papers
Live HTTP checks of every generated DOI
Builds an APA 7th .docx report with charts via python-docx
Saves summaries to Obsidian and uploads cited references to NotebookLM

When to use it

Producing a literature review with verified citations for a research topic
Mapping a field across several academic databases in one run
Building a NotebookLM source set from the references of a review

Who it is for: Researchers and students who want an agent to draft a verified, citation-checked literature review.

How it fits with Hermes Agent

It is built as a Hermes skill (the repository carries the hermes-skill topic) and is designed for the Hermes/ECC framework, with compatible ECC or Claude Code runners also named.

How to install Deep Research Agent

These commands are copied from the project's README. Check the repository for the latest steps before you run them.

pip install python-docx PyMuPDF requests matplotlib seaborn pandas
git clone https://github.com/CYC2002tommy/Deep-Research-Agent.git

Requirements: Hermes Agent or a compatible ECC/Claude Code runner, Node.js (v18+ recommended), Python 3.10+, a free Elsevier Scopus API key, and one-time NotebookLM MCP authentication

Note: The README assumes Windows-style output and Obsidian paths and recommends a university network for full-text PDF retrieval.

FAQ

What is Deep Research Agent?

Deep Research Agent is an agent skill that automates an academic literature review in seven phases. It searches several databases, verifies DOIs, runs a peer-review pass and produces a Word document.

Does Deep Research Agent work with Hermes Agent?

Yes, it is designed for the Hermes/ECC framework and is cloned into the agent's skills directory, such as ~/.hermes/skills/. The README also names compatible ECC or Claude Code runners.

What do I need to run Deep Research Agent?

You need a Hermes-compatible runner, Node.js, Python 3.10 or newer, a free Elsevier Scopus API key and NotebookLM authentication. The Exa Search MCP server is highly recommended.