Summary

The Deep Research Agent is an autonomous AI pipeline designed to automate the academic literature review process. It targets researchers and power users by providing rigorous, end-to-end automation from background sourcing and multi-database retrieval to full-text verification, quality-controlled drafting, and knowledge base ingestion, aiming to significantly reduce the manual effort in academic research.

How It Works

The agent employs a sophisticated 7-phase architecture. It begins with a planning phase and background scraping of high-impact papers. Phase 1 involves concurrent multi-agent discovery across Scopus, Exa, OpenAlex, and Semantic Scholar, strictly adhering to Q1-Q2 journal quality standards. Subsequent phases focus on deep extraction of metadata and full texts, structural drafting with evidence-backed claims, and a critical "Zero-Hallucination Guarantee" that mandates full-text verification and DOI resolution checks. An integrated "Remi Review" skill enforces academic tone by stripping AI-generated fluff, followed by compilation into APA 7th formatted .docx documents and ingestion into Obsidian and NotebookLM knowledge bases.

Quick Start & Requirements

• Installation: Clone the repository into your agent's skills directory (e.g., git clone https://github.com/CYC2002tommy/Deep-Research-Agent.git).
• Prerequisites:
  • Hermes Agent (or compatible ECC/Claude Code runner).
  • Node.js (v18+ recommended) and npx.
  • Python 3.10+.
  • Python dependencies: pip install python-docx PyMuPDF requests matplotlib seaborn pandas.
  • Required MCP Servers: scopus-mcp (requires free Elsevier Scopus API Key SCOPUS_API_KEY), notebooklm-mcp-server (requires npx notebooklm-mcp-server auth), Exa Search MCP (highly recommended), GitHub MCP & Playwright MCP (optional).
  • Local Environment: Obsidian vault path configured (defaulting to %OBSIDIAN_VAULT_PATH%\Hermes\ or %USERPROFILE%\Documents\Obsidian Vault\Hermes\). Output drive configured (defaulting to D:\Tommy).
  • Network: University/Academic network recommended for bypassing paywalls and ensuring full-text PDF access via cloakbrowser.

Highlighted Details

• Precision Sourcing: Two-stage pipeline with abstract screening followed by deep full-text reading (Methodology, Results) for highly relevant subsets.
• Strict Quality Control: Targets Q1-Q2 journals exclusively, banning Q4 and MDPI publications.
• Zero-Hallucination Guarantee: Mandates live HTTP requests to verify DOI validity and cross-references claims against raw full texts.
• Automated Outputs: Generates fully formatted APA 7th .docx documents with auto-generated data visualizations and ingests summaries/references into Obsidian and Google NotebookLM.

Maintenance & Community

No specific details on contributors, sponsorships, or community channels (Discord/Slack) were found in the provided README text.

Licensing & Compatibility

• License: MIT.
• Compatibility: The MIT license generally permits commercial use and linking with closed-source projects.

Limitations & Caveats

The agent is an AI assistant, not an author; human researchers assume 100% responsibility for accuracy, validity, and originality, requiring critical review and personal verification of all claims and cited literature. Transparency and disclosure of AI usage are mandatory per journal/institutional policies. Full-text PDF access for restricted databases heavily relies on network access via a university or academic institution.