mosaic-mcp
Pre-clinical drug discovery intelligence MCP server providing 44 tools to query 800+ drug targets, 12K+ compounds, 46K+ papers, 18K+ clinical trials, and 16K+ patents.
README
mosaic-mcp
<!-- mcp-name: io.github.sourabhnk/mosaic-mcp -->
Pre-clinical drug discovery intelligence as an MCP server. Query 800+ drug targets, 12K+ compounds, 46K+ papers, 18K+ clinical trials, and 16K+ patents through 44 specialized tools — 16 free for discovery, 28 Pro for competitive landscapes, whitespace, and thesis-grade analysis.
Quick Start
pip install mosaic-mcp
With Claude Desktop
Add to your claude_desktop_config.json:
{
"mcpServers": {
"mosaic": {
"command": "mosaic-mcp",
"env": {
"DATABASE_URL": "postgresql://reader:password@ep-xxx-pooler.neon.tech/mosaic_db?sslmode=require"
}
}
}
}
With Claude Code
claude mcp add mosaic -- mosaic-mcp
Standalone
export DATABASE_URL="postgresql://..."
export MOSAIC_API_KEY="msk_..." # Optional — for Pro tools
mosaic-mcp # stdio transport (default)
mosaic-mcp --transport sse --port 3001 # SSE for remote clients
Tools
Free Tier (16 tools) — discovery + workspace
| Tool | Description |
|---|---|
mosaic_search_targets |
Search drug targets by name, gene symbol, or keyword |
mosaic_get_target_profile |
Comprehensive target dossier (biology, compounds, scores) |
mosaic_get_target_compounds |
Compounds tested against a target with SAR data |
mosaic_get_target_patents |
Patent landscape for a target |
mosaic_get_target_papers |
Literature for a target |
mosaic_get_target_structure |
3D structure and ligandability summary for a target |
mosaic_kg_stats |
Knowledge graph overview statistics |
mosaic_list_indications |
Available disease indications |
mosaic_list_subindications |
Sub-indications within an indication area |
mosaic_subindication_breakdown |
Per-target activity across sub-indications |
mosaic_target_scores |
Target attractiveness scoring |
mosaic_target_wishlist_add |
Add a target to your personal wishlist |
mosaic_watchlist_create |
Create a watchlist |
mosaic_watchlist_add_item |
Add an item to a watchlist |
mosaic_watchlist_get |
Retrieve a watchlist |
mosaic_watchlist_list |
List your watchlists |
Pro Tier (28 additional tools) — analysis + whitespace
The committed-verdict layer. Includes the whitespace and differentiation tools:
mosaic_synthetic_lethal_whitespace, mosaic_modality_gaps,
mosaic_resistance_bypass_map, mosaic_find_undruggable_targets,
mosaic_talent_migration, mosaic_emerging_signals, and
mosaic_assess_druggability.
Plus the full analysis set: competitive landscape, pathway context, compound selectivity, indication landscape, target validation, clinical pipeline, compound analogs, target comparison, similar-target search, opportunity finding, organization portfolio, target network, mechanism of action, evidence maps, relation search, polypharmacology, clinical trial results, FDA regulatory status, drug comparison, drug repurposing candidates, and KOL discovery.
Data Coverage
| Entity | Count |
|---|---|
| Drug Targets | 801 (oncology + neuroscience + cardiovascular) |
| Compounds | 12,756 |
| Clinical Trials | 18,324 |
| Papers | 46,354 |
| Patents | 16,168 |
| Semantic Relations | 13,704 |
| Indications | 15,013 |
| Organizations | 45,496 |
Configuration
| Variable | Required | Description |
|---|---|---|
DATABASE_URL |
Yes | PostgreSQL connection string (Neon hosted or self-hosted) |
MOSAIC_API_KEY |
No | API key for Pro tool access |
MOSAIC_TIER |
No | Override tier (free, pro, enterprise) |
License
Apache 2.0 for the MCP server tools. The hosted knowledge graph data requires a subscription for Pro-tier access.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。
mcp-server-qdrant
这个仓库展示了如何为向量搜索引擎 Qdrant 创建一个 MCP (Managed Control Plane) 服务器的示例。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。