med-paper-assistant 0.6.3

A medical paper writing assistant using MCP support

Medical Paper Assistant

🔬 An Integrated AI Toolkit for Medical Paper Writing

3 MCP Servers · ~138 Tools · 26 Skills · 15 Prompt Workflows — All in VS Code

📖 繁體中文版 🤖 Auto-Paper: Fully Autonomous Paper Writing Guide — 11-Phase Pipeline, 78 Quality Hooks, Structured Review Loop

---

📦 What's in the Box

This is a monorepo toolkit that bundles everything a medical researcher needs — from literature search to Word/LaTeX export — into one integrated VS Code environment.

Component	Type	Tools	Description
mdpaper	Core MCP Server	88	Paper writing: 88 tools, plus 3 MCP prompts and 3 MCP resources
pubmed-search	MCP Server (submodule)	37	PubMed/Europe PMC/CORE search, PICO, citation metrics, session mgmt
CGU	MCP Server (submodule)	13	Creative generation: brainstorm, deep think, spark collision
VS Code Extension	Extension	5 cmds + 10 chat	MCP auto-registration, workspace setup, @mdpaper chat participant
Dashboard	Next.js Web App	—	Project management UI, diagram editor
Foam	VS Code Extension	—	[[wikilink]] citation linking, hover preview, graph view
Skills	Agent Workflows	26	Guided multi-tool workflows (literature review, draft writing...)
Prompts	Prompt Files	15	/mdpaper.search, /mdpaper.draft, etc.

External MCP Servers (optional, installed via uvx):

• drawio — CONSORT/PRISMA flowchart generation
• zotero-keeper — Import references from Zotero library

VSX note: The MedPaper VS Code extension installs Python MCP tools persistently per machine via uv tool install, attempts uv tool upgrade on later activations, and skips duplicate PubMed Search / Zotero Keeper registration when another installed VS Code extension already provides those MCP servers. CI smoke now covers ubuntu-latest, windows-latest, macos-13, and macos-14, including official MCP client checks plus VSX validation smoke.

How the Pieces Fit Together

flowchart LR
    subgraph IDE["VS Code"]
        Agent["Copilot Agent<br/>26 Skills · 15 Prompts"]
        Foam[Foam Plugin]
        Ext[MedPaper Extension]
        Dash[Dashboard]
    end

    subgraph MCP["MCP Servers (~138 tools)"]
      mdpaper["mdpaper<br/>88 tools + 3 prompts + 3 resources<br/>Draft · Export · Validate · Review"]
        pubmed["pubmed-search<br/>37 tools<br/>Search · Metrics"]
        cgu["CGU<br/>13 tools<br/>Deep Think · Ideas"]
    end

    subgraph Data["Project Data"]
        proj[("projects/{slug}/<br/>· .memory/<br/>· references/<br/>· drafts/")]
    end

    Agent <-->|MCP| mdpaper
    Agent <-->|MCP| pubmed
    Agent <-->|MCP| cgu
    mdpaper -->|HTTP API| pubmed
    Foam <-->|Wikilinks| proj
    mdpaper <--> proj
    Ext --> mdpaper
    Dash --> proj

---

🎯 Why This Tool?

Traditional paper writing tools require you to know exactly what you want before you start. But research is rarely that linear.

Medical Paper Assistant is different:

• 🔍 Explore First, Decide Later — Browse literature freely, save interesting papers, then decide your research direction
• 💬 Conversational Workflow — Chat naturally with AI to refine your ideas, not fight with forms
• 🧭 Guided Process — Step-by-step prompts guide you from concept to publication-ready manuscript
• 🔗 All-in-One — Search, write, cite, analyze, export — all integrated inside VS Code

Traditional Tools	Medical Paper Assistant
Fixed templates, rigid workflow	Flexible, exploratory approach
Separate apps for search/write/cite	All-in-one: ~138 tools in VS Code
Manual reference management	Auto-save with verified PubMed data
Export then format	Direct Word export with journal styles
Learn complex UI	Natural language conversation

---

🚀 Quick Start

Prerequisites

Requirement	Version	Check
Python	3.12+	python3 --version
Git	Any recent	git --version
VS Code	Latest	Help → About
GitHub Copilot	Extension	Extensions panel

Install

# Clone with submodules
git clone --recursive https://github.com/u9401066/med-paper-assistant.git
cd med-paper-assistant

# Run setup script
./scripts/setup.sh          # Linux/macOS
.\scripts\setup.ps1         # Windows PowerShell

The script will:

1. ✅ Create Python virtual environment (.venv/)
2. ✅ Initialize pinned Git submodules from this repository
3. ✅ Install all dependencies (via uv)
4. ✅ Create .vscode/mcp.json configuration for mdpaper, pubmed-search, cgu, zotero-keeper, asset-aware, and drawio
5. ✅ Verify MedPaper and CGU startup paths

Important installation notes:

• The setup script uses pinned submodule commits for reproducible installs. It does not auto-track the latest upstream submodule HEAD.
• If you intentionally want newer submodule code, run git submodule update --remote --merge yourself and test before committing.
• drawio uses npx -y @drawio/mcp, so Node.js/npm must be available on the machine.
• zotero-keeper and pubmed-search are launched via uvx in the repo workflow. In the VSX workflow they may be provided either by MedPaper itself or by separate installed VS Code extensions.

Verify: In Copilot Chat, type /mcp — you should see mdpaper listed 🎉

Optional Integrations

# Foam for reference linking (highly recommended)
code --install-extension foam.foam-vscode

# Draw.io for diagram generation
./scripts/setup-integrations.sh && ./scripts/start-drawio.sh

Windows PowerShell:

.\scripts\setup-integrations.ps1
.\scripts\start-drawio.ps1

---

💬 MCP Prompts — Just Type and Go

In Copilot Chat, type these prompts to trigger guided workflows:

Prompt	Description
/mdpaper.search	🔍 Start here! Explore literature, save papers
/mdpaper.concept	📝 Develop research concept with novelty validation
/mdpaper.draft	✍️ Write manuscript with auto-citations
/mdpaper.analysis	📊 Analyze CSV data, generate figures & Table 1
/mdpaper.format	📄 Export to Word with journal formatting
/mdpaper.clarify	🔄 Refine specific sections through conversation
/mdpaper.project	📁 Create or switch research projects
/mdpaper.strategy	⚙️ Configure search strategy (dates, filters)
/mdpaper.help	❓ Show all available commands

💡 Recommended Workflow: /mdpaper.search → /mdpaper.concept → /mdpaper.draft → /mdpaper.format

---

🧠 Skill System + Project Memory

Our core differentiator: We don't just provide tools — we provide guided workflows that know how to combine tools effectively, AND project memory that remembers your research journey across sessions.

What is a Skill?

Tool  = Single capability (search, save, analyze...)
Skill = Complete knowledge (how to combine tools to accomplish tasks)

26 Skills covering the full research lifecycle:

Category	Skills	Triggers
🔬 Research	literature-review, concept-development, concept-validation, parallel-search	"找論文", "search", "concept", "validate"
✍️ Writing	draft-writing, reference-management, word-export	"寫草稿", "draft", "citation", "export"
📁 Management	project-management, memory-updater, memory-checkpoint	"新專案", "切換", "存檔"
🛠️ Development	git-precommit, code-refactor, test-generator, code-reviewer	"commit", "refactor", "test"

Project Memory

Each project maintains its own .memory/ folder, so the AI continues previous research coherently:

projects/{slug}/
├── .memory/
│   ├── activeContext.md   ← Agent's working memory
│   └── progress.md        ← Research milestones
├── concept.md             ← Research concept (with 🔒 protected sections)
├── references/            ← Foam-compatible literature library
├── drafts/                ← Markdown drafts with [[citations]]
├── data/                  ← CSV data files
└── results/               ← Figures, .docx exports

---

✨ Key Features

Literature & References

• PubMed + Europe PMC + CORE search (37 search tools)
• PICO parsing for clinical questions
• MCP-to-MCP verified data — PMID sent directly, no agent hallucination
• Layered trust: 🔒 VERIFIED (PubMed) · 🤖 AGENT (AI notes) · ✏️ USER (your notes)
• Foam wikilinks: [[author2024_12345678]] with hover preview & backlinks

Writing & Editing

• AI draft generation per section (Introduction, Methods, Results, Discussion)
• Citation-Aware Editing — patch_draft validates all [[wikilinks]] before saving
• Auto-fix citation format — [[12345678]] → [[author2024_12345678]]
• Novelty validation — 3-round independent scoring (threshold: 75/100)
• Anti-AI writing rules — Evidence funnel structure, no clichés

Data Analysis

• CSV dataset analysis with descriptive statistics
• Statistical tests (t-test, ANOVA, chi², correlation, Mann-Whitney, Fisher's)
• Table 1 generator — Baseline characteristics with automatic variable detection
• Publication-ready figures (matplotlib/seaborn)

Export & Submission

• Word export with journal template support
• Cover letter + highlights generation
• Manuscript consistency checker
• Reviewer response generator (point-by-point format)
• Submission checklist (word count, figure format, etc.)

Infrastructure

• DDD Architecture (Domain-Driven Design) with clean layer separation
• 15 pre-commit hooks (ruff, mypy, bandit, pytest, prettier, doc-update...)
• Workspace State recovery for cross-session continuity
• uv for all Python package management
• MCP SDK features in active use — tools, elicitation, and progress notifications for long-running audit/review operations

---

🏗️ Architecture

┌──────────────────────────────────────────────────────────────────────────┐
│                          👤 User Layer                                    │
│  ┌─────────────────┐    ┌──────────────────────────────┐  ┌──────────┐  │
│  │   VS Code        │    │  Foam Extension               │  │Dashboard │  │
│  │   Editor         │    │  [[wikilinks]] autocomplete    │  │(Next.js) │  │
│  │                  │    │  hover preview · backlinks     │  │          │  │
│  └─────────────────┘    └──────────────────────────────┘  └──────────┘  │
└──────────────────────────────────────────────────────────────────────────┘
                                    │
                                    ▼
┌──────────────────────────────────────────────────────────────────────────┐
│               🤖 Copilot Agent (Orchestrator)                             │
│      26 Skills + 15 Prompt Workflows + Agent Customization               │
│   /mdpaper.search → /mdpaper.concept → /mdpaper.draft → export          │
└───────┬──────────────────┬──────────────────┬──────────────────┬─────────┘
        │                  │                  │                  │
        ▼                  ▼                  ▼                  ▼
┌───────────────┐  ┌───────────────┐  ┌───────────────┐  ┌───────────────┐
│ 📝 mdpaper    │  │🔍 pubmed-     │  │💡 cgu         │  │🔌 External    │
│  88 tools     │  │  search       │  │  13 tools     │  │   MCPs (uvx)  │
│               │  │  37 tools     │  │               │  │               │
│ • projects    │  │ • PubMed      │  │ • brainstorm  │  │ 🎨 drawio     │
│ • references  │  │ • Europe PMC  │  │ • deep_think  │  │ • diagrams    │
│ • drafts      │  │ • CORE        │  │ • spark       │  │               │
│ • validation  │  │ • PICO        │  │ • methods     │  │ 📖 zotero     │
│ • analysis    │  │ • Gene/Chem   │  │               │  │ • import refs │
│ • export      │  │ • Session     │  │               │  │               │
└───────┬───────┘  └───────────────┘  └───────────────┘  └───────────────┘
        │
        ▼
┌──────────────────────────────────────────────────────────────────────────┐
│                          💾 Local Storage                                 │
│  projects/{slug}/                                                        │
│  ├── concept.md          ← Research concept with 🔒 protected sections   │
│  ├── references/{pmid}/  ← Foam-compatible .md + metadata.json           │
│  ├── drafts/             ← Markdown drafts with [[citations]]            │
│  ├── data/               ← CSV data files                                │
│  └── results/            ← Figures, .docx exports                        │
└──────────────────────────────────────────────────────────────────────────┘

MCP-to-MCP Direct Communication

When saving references, data flows directly between MCP servers — the Agent only passes a PMID, never full metadata:

Agent: "save PMID:24891204"
     │
     ▼
mdpaper.save_reference_mcp(pmid="24891204")
     │  Direct HTTP call (not through Agent)
     ▼
pubmed-search: GET /api/cached_article/24891204
     │  Returns verified PubMed data
     ▼
Saved with layered trust:
  🔒 VERIFIED: PubMed data (immutable)
  🤖 AGENT:    AI notes (marked source)
  ✏️ USER:     Your notes (editable)

---

🛠️ mdpaper MCP Tools

88 active tools organized into 7 groups, plus 3 MCP prompts and 3 MCP resources for official MCP clients.

📁 Project Management (17 tools)

Projects, exploration mode, workspace state recovery, diagram management.

Key Tools	Description
create_project / switch_project / delete_project	Project lifecycle
start_exploration / convert_exploration_to_project	Explore-first workflow
get_workspace_state / sync_workspace_state	Cross-session recovery
save_diagram / list_diagrams	Draw.io integration
setup_project_interactive	Interactive paper type configuration
update_authors	Manage structured author metadata

📚 Reference Management (12 tools)

Save, search, format, and manage references with Foam integration.

Key Tools	Description
save_reference_mcp	Recommended — Save by PMID via MCP-to-MCP (verified data)
list_saved_references / search_local_references	Browse & search library
format_references / set_citation_style	Vancouver / APA / Nature
sync_references	Sync [[wikilinks]] to numbered references

✍️ Draft & Editing (13 tools)

Write, edit, cite — with built-in validation.

Key Tools	Description
draft_section / write_draft	Create and write sections
list_drafts / read_draft / delete_draft	Draft lifecycle
get_available_citations	List all valid [[citation_key]] before editing
patch_draft	Citation-aware partial edit with wikilink validation
insert_citation / suggest_citations	Smart citation insertion
scan_draft_citations / sync_references	Citation management
count_words	Section and manuscript word-count checks

✅ Validation (3 tools)

Tool	Description
validate_concept	Full novelty scoring against the active concept
validate_wikilinks	Auto-fix [[12345678]] → [[author2024_12345678]]
compare_with_literature	Compare the current idea against saved references

📊 Data Analysis (10 tools)

Tool	Description
analyze_dataset	Descriptive statistics for CSV
run_statistical_test	t-test, ANOVA, chi², correlation, etc.
generate_table_one	Baseline characteristics with auto variable detection
create_plot	Publication-ready figures
insert_figure	Insert figure into draft with archive validation
insert_table	Insert table into draft with archive validation
list_assets	List figures and tables in project results

🔍 Review & Audit (23 tools)

Category	Key Tools
Pipeline Gates	validate_phase_gate, pipeline_heartbeat, validate_project_structure
Review Loop	start_review_round, submit_review_round, request_section_rewrite
Pipeline Control	pause_pipeline, resume_pipeline, approve_section
Audit & Hooks	run_quality_audit, run_writing_hooks, record_hook_event
Self-Evolution	run_meta_learning, verify_evolution, apply_pending_evolutions
Domain Constraints	check_domain_constraints, evolve_constraint
Data & Health	validate_data_artifacts, diagnose_tool_health, check_formatting

📄 Export & Submission (10 tools)

Category	Key Tools
Word Export	export_word, list_templates, start_document_session, verify_document
Pandoc Export	export_docx, export_pdf, preview_citations, build_bibliography
Submission	generate_cover_letter, generate_highlights

🧩 MCP Prompts & Resources

Capability	Names / URIs	Purpose
Prompts	project_bootstrap, draft_section_plan, word_export_checklist	Materialize guided prompt workflows through the official MCP prompt API
Resources	medpaper://workspace/state, medpaper://workspace/projects, medpaper://templates/catalog	Surface live workspace state, project lists, and template metadata via MCP resources

🔍 pubmed-search MCP Tools (37 tools)

Category	Key Tools
Search	search_literature, generate_search_queries, parse_pico
Databases	PubMed, Europe PMC (fulltext + text mining), CORE (200M+ open access)
Gene/Chem	search_gene, get_gene_details, search_compound, search_clinvar
Exploration	find_related_articles, find_citing_articles, get_article_references
Export	prepare_export (RIS/BibTeX/CSV), get_citation_metrics (iCite RCR)
Session	get_session_pmids, list_search_history (survives AI memory limits)

💡 CGU Creative Tools (13 tools)

Category	Key Tools
Ideation	generate_ideas, spark_collision, spark_collision_deep
Analysis	deep_think, multi_agent_brainstorm
Methods	list_methods, select_method, apply_method

CGU runtime notes:

• In the repository workflow, CGU is started from the pinned submodule with uv run --directory integrations/cgu python -m cgu.server.
• In the VSX workflow, MedPaper registers CGU when bundled code or the workspace submodule is available; otherwise CGU is simply skipped.
• CGU itself supports Python >=3.11, but this repository currently requires Python >=3.12, so cross-platform repo setup should be treated as Python 3.12 baseline on macOS, Linux, and Windows.
• The default repo MCP example uses CGU_THINKING_ENGINE=simple, which is the low-friction mode. Advanced LLM-backed modes still depend on CGU-side model/provider configuration.

---

🔗 Foam Integration

Feature	How to Use	Benefit
Wikilinks	[[greer2017_27345583]]	Link references in drafts
Hover Preview	Mouse over any [[link]]	See abstract without opening file
Backlinks Panel	Open reference file	See which drafts cite this paper
Graph View	Ctrl+Shift+P → Foam: Show Graph	Visualize paper connections
Project Isolation	Auto-switches on switch_project	Only see current project's references

Citation Autocomplete

Type [[ in any draft to trigger the autocomplete menu:

According to previous studies [[    ← Type [[ here
                              ┌─────────────────────────────┐
                              │ 🔍 greer2017_27345583       │
                              │    smith2020_12345678       │
                              │    chen2019_87654321        │
                              └─────────────────────────────┘

Search by author ([[greer), year ([[2017), PMID ([[27345583), or keyword ([[sedation).

---

📚 Reference File Structure

References are stored with Foam-optimized, layered-trust structure:

references/{pmid}/
├── {citation_key}.md   ← YAML frontmatter + abstract (human-readable)
└── metadata.json       ← Full metadata (programmatic access)

---
# 🔒 VERIFIED (from PubMed, immutable)
title: "Complications of airway management"
author:
  - { family: Pacheco-Lopez, given: Paulette C }
year: 2014
journal: Respiratory Care
pmid: "24891204"
_source:
  mcp: pubmed-search
  verified: true

# 🤖 AGENT (AI-generated, marked)
_agent:
  notes: "Key review on airway complications"
  relevance: high

# Foam
aliases: [pachecolopez2014, "PMID:24891204"]
tags: [reference, airway, review]
---

---

📂 Project Structure

med-paper-assistant/
├── src/med_paper_assistant/       # Core MCP server (DDD architecture)
│   ├── domain/                    #   Business logic, entities, value objects
│   ├── application/               #   Use cases, services
│   ├── infrastructure/            #   DAL, external services
│   └── interfaces/mcp/            #   MCP server, 88 tools + 3 prompts + 3 resources
│
├── integrations/                  # Bundled MCP servers
│   ├── pubmed-search-mcp/         #   PubMed/PMC/CORE search (37 tools)
│   └── cgu/                       #   Creative generation (13 tools)
│
├── vscode-extension/              # VS Code Extension
│   ├── src/                       #   Extension source
│   ├── skills/                    #   Agent skill definitions
│   └── prompts/                   #   Quick-action prompts
│
├── dashboard/                     # Next.js project management UI
│   └── src/
│
├── projects/                      # Research projects (isolated workspaces)
│   └── {slug}/
│       ├── .memory/               #   Cross-session AI memory
│       ├── concept.md             #   Research concept
│       ├── references/            #   Local reference library
│       ├── drafts/                #   Markdown drafts
│       └── results/               #   Figures, exports
│
├── .claude/skills/                # 26 Agent skill definitions
├── .github/prompts/               # 15 Prompt workflow files
├── templates/                     # Journal Word templates
├── memory-bank/                   # Global project memory
└── tests/                         # pytest test suite

---

🗺️ Roadmap

Status	Feature	Description
✅	3 MCP Servers	mdpaper (88) + pubmed-search (37) + CGU (13)
✅	Foam Integration	Wikilinks, hover preview, backlinks, project isolation
✅	Project Memory	.memory/ for cross-session AI context
✅	Table 1 Generator	Auto-generate baseline characteristics
✅	Novelty Validation	3-round scoring with 75/100 threshold
✅	Citation-Aware Editing	patch_draft with wikilink validation
✅	MCP-to-MCP Trust	Verified PubMed data via direct HTTP
✅	Pre-commit Hooks	15 hooks (ruff, mypy, bandit, pytest, prettier...)
🔜	Full VSX Extension	TreeView, CodeLens, Diagnostics (Direction C)
🔜	Pandoc Export	Word + LaTeX dual export with CSL citations
📋	Systematic Review	PRISMA flow, Risk of Bias, meta-analysis
📋	AI Writing Intelligence	Citation intelligence, coherence engine
📋	REST API Mode	Expose tools as REST API

Architecture Direction: Direction C — Full VSX + Foam + Pandoc

Legend: ✅ Complete | 🔜 In Progress | 📋 Planned

---

🤝 Contributing

We welcome contributions! See CONTRIBUTING.md for guidelines.

• 🐛 Report bugs — Open an issue
• 💡 Suggest features — Share your ideas
• 🔧 Submit code — Fork → Branch → PR

---

� Citation

If you use Medical Paper Assistant in your research, please cite our paper:

@article{medpaper2025,
  title   = {MedPaper Assistant: A Self-Evolving, MCP-Based Framework for
             AI-Assisted Medical Paper Writing with Closed-Loop Quality Assurance},
  author  = {[Authors]},
  year    = {2025},
  note    = {Submitted to medRxiv},
  url     = {https://github.com/u9401066/med-paper-assistant}
}

Note: This paper was produced entirely by the MedPaper Assistant's autonomous pipeline as a self-referential demonstration. The manuscript, audit trail, and all quality metrics are available in projects/self-evolving-ai-paper-writing-framework/. The preprint is being submitted to medRxiv — this section will be updated with the DOI once available.

---

�📄 License

Apache License 2.0 — See LICENSE