Content
# Patent Space MCP
Patent intelligence MCP server with 38 tools covering 13.6M+ Japanese patents, 607 technology clusters, and 4,300+ resolved entities.
## Features
- **13.6M+ Japanese patents** with full metadata, CPC codes, and citation data
- **607 technology clusters** derived from CPC classification
- **4,300+ company entities** with name resolution (Japanese/English/ticker)
- **Startability engine** — measures firm readiness to enter technology areas
- **Patent finance tools** — Black-Scholes option valuation, VaR, CAPM beta
- **Network analysis** — citation topology, knowledge flow, resilience testing
- **GDELT integration** — media sentiment signals for ~46 major firms
- **NVMe-optimized** — fast-path queries return in <1s
## Quick Start
### As Claude Desktop MCP Server
Add to your Claude Desktop config (`claude_desktop_config.json`):
```json
{
"mcpServers": {
"patent-space": {
"url": "https://patent-space.dev/mcp"
}
}
}
```
### As Smithery MCP
Available on [Smithery](https://smithery.ai/) as `patent-space-mcp`.
## Tools (38)
### Patent Search & Detail (2)
| Tool | Description |
|------|-------------|
| `patent_search` | Search patents by keyword, CPC code, applicant, or date range |
| `patent_detail` | Get full patent record including CPC codes, assignees, inventors, and citations |
### Company Analysis (5)
| Tool | Description |
|------|-------------|
| `firm_patent_portfolio` | Patent portfolio analysis — counts, CPC distribution, filing trends, co-applicants |
| `patent_compare` | Compare multiple firms by patent volume, CPC mix, and trend |
| `applicant_network` | Build co-applicant network for a target applicant |
| `similar_firms` | Discover firms with similar patent portfolios via cosine similarity |
| `firm_tech_vector` | Get a firm's precomputed technology vector and diversity metadata |
### Startability — Market Entry Analysis (4)
| Tool | Description |
|------|-------------|
| `startability` | Compute startability score for a firm-technology pair |
| `startability_delta` | Compute change in startability scores over time |
| `startability_ranking` | Rank tech clusters or firms using precomputed startability surface |
| `tech_fit` | Compute phi_tech fit components for a firm and technology cluster |
### Technology Analysis (7)
| Tool | Description |
|------|-------------|
| `tech_landscape` | Analyze filing trends and top applicants in a technology area |
| `tech_clusters_list` | List 607 technology clusters with optional CPC filtering |
| `tech_trend` | Analyze time-series technology trends with growth rates and new entrants |
| `tech_trend_alert` | Detect hot and cooling technology trends with market signals |
| `tech_entropy` | Technology maturity via Shannon entropy of applicant diversity |
| `tech_volatility` | Technology volatility analysis with decay curve and half-life |
| `tech_beta` | CAPM-style technology beta: market sensitivity analysis |
### Network Analysis (4)
| Tool | Description |
|------|-------------|
| `citation_network` | Build a patent citation network around a patent or firm's top patents |
| `network_topology` | Citation network topology analysis (scale-free, small-world, clustering) |
| `network_resilience` | Patent network resilience via percolation theory |
| `knowledge_flow` | Cross-CPC knowledge flow analysis via citation patterns |
### Strategy & M&A (4)
| Tool | Description |
|------|-------------|
| `adversarial_strategy` | Game-theoretic analysis of two firms' technology territories |
| `ma_target` | Recommend M&A acquisition targets based on patent portfolio analysis |
| `tech_gap` | Analyze technology gap and synergy between two firms |
| `sales_prospect` | Identify and rank patent licensing sales targets |
### Invention & FTO Analysis (3)
| Tool | Description |
|------|-------------|
| `invention_intelligence` | Analyze technology description for prior art, FTO risk, and whitespace |
| `cross_domain_discovery` | Discover cross-domain technology clusters related to a query |
| `cross_border_similarity` | Detect similar patent filings across international jurisdictions |
### Patent Finance & Valuation (4)
| Tool | Description |
|------|-------------|
| `patent_market_fusion` | Combine patent portfolio strength with market signals |
| `patent_valuation` | Score patent, portfolio, or technology area value with royalty rate reference |
| `patent_option_value` | Black-Scholes real option valuation for patents |
| `portfolio_var` | Portfolio Value-at-Risk for patent expiration risk |
### Technology Fusion & Evolution (3)
| Tool | Description |
|------|-------------|
| `tech_fusion_detector` | Technology fusion detector via co-citation analysis |
| `portfolio_evolution` | Track how a firm's technology portfolio evolved over time |
| `bayesian_scenario` | Bayesian patent investment simulation with data-driven priors |
### Utility (2)
| Tool | Description |
|------|-------------|
| `entity_resolve` | Resolve a company name to its canonical form |
| `gdelt_company_events` | Fetch GDELT events and cached five-axis features for a firm |
## Data Sources
| Source | Coverage | Size |
|--------|----------|------|
| Google Patents Public Data (BigQuery) | 13.6M+ Japanese patents | 310GB+ |
| CPC Classification | 607 technology clusters | Pre-computed |
| GDELT Global Knowledge Graph | Media signals for ~46 major firms (2020-2024) | Cached |
| Entity Registry | TSE Prime/Standard/Growth + S&P500 + Global | 4,300+ firms |
## Architecture
```
┌─────────────────────────────────────────────┐
│ Claude / MCP Client │
└─────────────┬───────────────────────────────┘
│ Streamable HTTP (/mcp)
┌─────────────▼───────────────────────────────┐
│ FastMCP 2.14.5 (Python 3.13) │
│ ┌──────────────────────────────────────┐ │
│ │ 38 Tool Functions + Entity Resolver │ │
│ └──────────────┬───────────────────────┘ │
│ ┌──────────────▼───────────────────────┐ │
│ │ PatentStore (SQLite, WAL mode) │ │
│ │ - patents: 13.6M rows │ │
│ │ - patent_cpc: 44.8M rows │ │
│ │ - startability_surface: 10.2M rows │ │
│ │ - firm_tech_vectors: 27.8K rows │ │
│ │ - tech_clusters: 607 rows │ │
│ └──────────────────────────────────────┘ │
│ Docker (48GB mem limit) │
│ NVMe SSD (2TB) │
└─────────────────────────────────────────────┘
```
## Performance
| Query Type | Response Time | Examples |
|-----------|--------------|---------|
| Pre-computed lookups | 3-40ms | entity_resolve, tech_fit, startability |
| Aggregation queries | 1-30s | tech_landscape, patent_compare |
| Full-text search | 100ms-5s | patent_search |
| Heavy analysis | 10-30s | invention_intelligence, network_topology |
## Self-Hosting
### Requirements
- 64GB+ RAM recommended
- 500GB+ NVMe SSD
- Docker + Docker Compose
### Setup
```bash
git clone https://github.com/Rei02061986/patent-space-mcp.git
cd patent-space-mcp
cp .env.example .env
# Configure PATENT_DB_PATH in .env
docker compose up -d
curl http://localhost:8001/health
```
## License
MIT
