medwriter-mcp

The Medical Writer's AI Toolkit — 33 expert prompts as callable MCP tools.

Built by a CMPP-certified medical writer with a PhD and 10+ years in pharma. Gives Claude and any MCP-compatible AI agent direct access to publication-ready prompt frameworks for every stage of the pharma medical writing workflow.

---

Free tier (5 tools) vs Pro tier (28 tools)

Tier	Tools	Access
Free	5 core tools	Install and use immediately
Pro	All 33 tools + ADAPT framework	See Pro tier below

---

Tools

✅ Free tier

Tool	Description
get_structured_abstract_prompt	Draft a CONSORT/STROBE-aligned structured journal abstract
get_manuscript_outline_prompt	Create a detailed manuscript outline with section word counts
get_peer_review_response_prompt	Draft a point-by-point reviewer response
get_author_invitation_email_prompt	Write an ICMJE-aligned author invitation email
get_linkedin_publication_post_prompt	Draft a LinkedIn post announcing a publication

🔐 Pro tier — 28 additional tools across 9 chapters

Chapter 1 — Abstract writing

• Convert structured → unstructured abstract
• Rewrite weak conclusions to be data-anchored
• Adapt abstracts for different audiences

Chapter 2 — Manuscript development

• Draft Introduction, Methods, Results, Discussion sections
• AMA style, CONSORT/STROBE aligned

Chapter 3 — Peer review

• Evidence-based rebuttal to reviewer disagreement
• Cover letter for revised manuscript submission

Chapter 4 — Congress abstracts and posters

• ASCO, ASH, ESMO, APA congress abstracts
• Declarative poster titles and take-home messages
• Timed oral presentation scripts

Chapter 5 — Slide deck narratives

• Slide-by-slide deck outlines by audience type
• Speaker notes for data slides
• Mechanism of action slide copy

Chapter 6 — Publication planning

• Publication plan framework table
• Literature and data gap analysis
• Executive summary for medical affairs leadership
• Journal selection with rationale (top 3 journals)

Chapter 7 — Author collaboration

• Author review request with deadline
• Response to author major revision request

Chapter 8 — Editing, QC and style

• Comprehensive AMA-style manuscript edit
• Promotional language check for MLR pre-review
• Statistical reporting consistency check

Chapter 9 — Career and professional development

• CV tailoring for ATS keyword alignment
• CMPP exam practice question generator
• Medical writing interview prep with STAR frameworks

Bonus — ADAPT framework

• Layer Audience, Document type, Accuracy level, Purpose, and Tone onto any prompt in the library

---

Quickstart

Install

git clone https://github.com/pubspro/medwriter-mcp.git
cd medwriter-mcp
pip install mcp

Add to Claude Desktop

Edit %APPDATA%\Claude\claude_desktop_config.json (Windows) or ~/Library/Application Support/Claude/claude_desktop_config.json (macOS):

{
  "mcpServers": {
    "medwriter-mcp": {
      "command": "python",
      "args": ["/absolute/path/to/medwriter-mcp/server.py"]
    }
  }
}

Restart Claude Desktop. All tools appear automatically.

Add to Claude Code

claude mcp add medwriter-mcp python /absolute/path/to/medwriter-mcp/server.py

---

Example prompts

Once connected, ask Claude:

Use get_structured_abstract_prompt to draft an abstract for a Phase 3 RCT 
of pembrolizumab in NSCLC. Primary endpoint: PFS HR 0.58 (95% CI 0.46–0.72, p<0.001).

Use get_gap_analysis_prompt to identify publication gaps for a BTK inhibitor 
in CLL based on the current literature landscape.

Use get_congress_abstract_prompt to draft an ASH abstract for a study 
of venetoclax plus obinutuzumab in previously untreated CLL.

Use get_promotional_language_check_prompt to review this slide deck excerpt 
before MLR submission.

Use apply_adapt_framework to rewrite this abstract for a nursing audience 
at a health literacy reading level.

---

Data safety

Every prompt is structured so you never need to input unpublished, confidential, or proprietary data. Tools marked 🔒 carry elevated data-sensitivity risk and include inline guidance.

Never input into any public AI tool:

• Unpublished clinical trial data
• Patient-level or identifiable data
• Proprietary compound information or IP
• Data under NDA or confidentiality agreement
• Pre-decisional regulatory documents

Always comply with ICMJE, GPP4, and your organization's AI use policies.

---

Who this is for

• Medical writers at associate through director level
• Publication managers and medical affairs professionals
• Pharmacists, physicians, and researchers moving into medical writing
• CMPP candidates preparing for certification

---

Companion servers

Pair with other pubspro MCP servers for a complete pharma AI workflow:

• pharma-mcp — ClinicalTrials.gov, PubMed, FDA, ICH guidelines
• pubmed-search — PubMed literature search
• medterms-mcp — ICD-10, MedDRA, RxNorm, CTCAE

---

Requirements

• Python 3.8+
• mcp (pip install mcp)

---

License

MIT