Drug Pipeline MCP Server

An MCP server that synthesizes pharmaceutical R&D pipeline intelligence — clinical trials, FDA/EMA approvals, safety signals, and labels — into single answers where every fact traces back to a primary source (NCT ID, FDA application number, or PMID).


Type	MCP server
Supplier	DasClown (community)
Availability	GA — published on PyPI (`drug-pipeline-mcp`)
Pricing	Free / OSS (MIT); all backing data sources are public and keyless
Capabilities	Read-only — aggregates public regulatory/trial/literature APIs
Verified	degraded · 2026-07-27 — repo resolves + git-source install works; PyPI drug-pipeline-mcp 404 so pip/uvx does not resolve
Security	caution · 2026-07-27 — provenance matches DasClown, MIT, no OSV advisories, single-maintainer + not on PyPI

How to install

No API key is needed — every backing source is publicly accessible. The server runs over stdio by default (Claude launches it on demand); an optional HTTP mode is available for remote use.

Prerequisite — Python 3.10+ and Claude Code / Claude Desktop.
Install from PyPI:
```
pip install drug-pipeline-mcp
```
(Or install the latest from source: pip install git+https://github.com/DasClown/drug-pipeline-mcp.git.)
Claude Code — register over stdio:
```
claude mcp add --transport stdio drug-pipeline -- uvx drug-pipeline-mcp
```
(uvx fetches and runs the package in an isolated env; if you installed with pip instead, use -- drug-pipeline in place of -- uvx drug-pipeline-mcp.)
Claude Desktop — add to claude_desktop_config.json:
```
{
  "mcpServers": {
    "drug-pipeline": {
      "command": "uvx",
      "args": ["drug-pipeline-mcp"]
    }
  }
}
```
(Equivalent to the Claude Code stdio registration; requires uv/uvx on your PATH. If you pip installed the package instead, use "command": "drug-pipeline" with empty "args".)
Claude Code / Desktop — optional HTTP mode (long-lived server; keep it running):
```
pip install "drug-pipeline-mcp[http]"
drug-pipeline --http --port 8081
```
Then claude mcp add --transport http drug-pipeline http://127.0.0.1:8081/mcp (Claude Code) or proxy via mcp-remote for Desktop.

What it does

Exposes six tools spanning trials, approvals, safety, and labels:

search_trials — clinical-trial discovery by condition/phase/status (ClinicalTrials.gov).
get_approvals — FDA approval history with submission dates (Drugs@FDA).
get_safety_data — FAERS adverse-event reports and reaction counts.
get_drug_label — FDA prescribing information and contraindications (Drug Labeling / DailyMed).
get_eu_approvals — EMA authorization status with orphan/biosimilar flags.
drug_pipeline — composite aggregation across all sources for a single drug.

Backing sources include ClinicalTrials.gov, openFDA (Drugs@FDA, FAERS, Labeling, NDC, Enforcement), RxNorm, PubMed/NCBI, EMA daily XLSX, DailyMed, Open Targets, and MyChem.info. The project makes no ML predictions — it only structures verified primary sources.

Primary use cases: competitive pipeline landscaping (e.g., “what’s in the pipeline for GLP-1 agonists?”), cross-region approval comparison (US vs EU), drug-safety signal review, repurposing-candidate scoping from trial/approval evidence.

Notes

Because outputs are aggregated from live public APIs, latency and completeness depend on those upstream services. The composite drug_pipeline tool fans out across all sources and can be slower than the single-source tools. This server overlaps but does not duplicate the discrete openFDA / OpenFDA MCP Server (cyanheads) and ClinicalTrials.gov MCP entries — its distinct value is the cross-source, source-traceable pipeline synthesis plus EMA approval coverage.

Sources

Installed this tool?

Share feedback — install path, OS, errors, workarounds. The form opens with this tool pre-selected and a link back to this page.