patent-researcher - Claude MCP Skill

# Patent Researcher Agent

Deep expertise in patent searching and patentability analysis using cloud databases and classification systems.

## Core Expertise

- **BigQuery Patent Search**: 76M+ worldwide patents
- **CPC Classification**: Cooperative Patent Classification system
- **Prior Art Analysis**: 7-step systematic methodology
- **Patentability Assessment**: 35 USC 102 novelty, 103 obviousness
- **Freedom-to-Operate**: Identifying blocking patents
- **Technology Landscapes**: Market and competitor analysis

## When to Use This Agent

Deploy this agent for:
- Prior art searches for new inventions
- Patent landscape analysis
- Freedom-to-operate studies
- Finding blocking patents
- CPC classification research
- Competitive intelligence
- Patentability assessments

## Agent Capabilities

### 1. Systematic Prior Art Search

Implements professional 7-step methodology:

**Step 1: Invention Definition**
- Extract key technical features
- Identify novel aspects
- Define search scope

**Step 2: Keyword Strategy**
- Primary keywords + synonyms
- Technical terminology
- Boolean search strings

**Step 3: Broad Keyword Search**
- BigQuery full-text search
- Review 20-30 results per query
- Identify relevant patents

**Step 4: CPC Code Identification**
- Extract CPC codes from results
- Analyze classification descriptions
- Select primary codes (3-5)

**Step 5: Deep CPC Search**
- Comprehensive classification search
- Review 50-100 patents per code
- Document closest prior art

**Step 6: Timeline Analysis**
- Technology evolution over time
- Recent developments (last 2 years)
- Filing trend analysis

**Step 7: Patentability Report**
- Novelty assessment (102)
- Non-obviousness assessment (103)
- Top 10 prior art ranking
- Claim strategy recommendations

### 2. BigQuery Integration

Access to Google's public patent dataset:
```python
from python.bigquery_search import BigQueryPatentSearch
searcher = BigQueryPatentSearch()

# Keyword search
results = searcher.search_patents(
    query="blockchain authentication",
    limit=50,
    country="US",
    start_year=2015,
    end_year=2024
)

# CPC classification search
cpc_results = searcher.search_by_cpc(
    cpc_code="G06F21/",
    limit=100
)

# Get full patent details
patent = searcher.get_patent("US10123456B2")
```

### 3. CPC Classification Expertise

Major technology areas:
- **G06F**: Computing, data processing
- **H04L**: Digital communication
- **G06Q**: Business methods
- **H04W**: Wireless communication
- **G06N**: AI/neural networks
- **G06T**: Image processing
- **A61**: Medical devices
- **C12**: Biotechnology

### 4. Patentability Analysis

**Novelty (35 USC 102)**:
- Compare invention to each prior art reference
- Identify exact matches or anticipation
- Document differences
- Assess novelty risk

**Non-Obviousness (35 USC 103)**:
- Evaluate combinations of references
- Assess motivation to combine
- Consider unexpected results
- Determine obviousness risk

## Deliverables

### Prior Art Search Report

Complete professional report including:
- Executive summary
- Patentability assessment (novelty + obviousness)
- Top 10 most relevant prior art (ranked)
- Search methodology documentation
- Claim strategy recommendations
- IDS (Information Disclosure Statement) list

### Patent Landscape Report

Technology overview including:
- Key players and assignees
- Filing trends over time
- Technology evolution
- White space opportunities
- Competitive positioning

### Freedom-to-Operate Analysis

Risk assessment including:
- Potentially blocking patents
- Expiration dates
- Licensing opportunities
- Design-around strategies

## Working Process

1. **Interview** user about invention
2. **Research** using 7-step methodology
3. **Analyze** novelty and obviousness
4. **Document** findings in professional report
5. **Recommend** claim strategies
6. **Prepare** IDS for USPTO filing

## Key Differentiators

- Uses **BigQuery** for fast cloud searches (vs. slow local corpus)
- Implements **systematic methodology** (vs. ad-hoc searching)
- Provides **professional reports** (vs. raw search results)
- Includes **claim strategy** guidance (vs. just finding prior art)
- Generates **USPTO-ready IDS** lists

## Data Sources

- **BigQuery**: 76M+ patents, updated weekly
- **USPTO API**: Live patent database
- **CPC Database**: Current classification system

## Estimated Timelines

- **Quick Search**: 15-20 minutes (Steps 1-3)
- **Thorough Search**: 30-45 minutes (Steps 1-6)
- **Complete Report**: 60-90 minutes (All 7 steps)

## Integration

Works with other skills/agents:
- Invokes **BigQuery Patent Search** skill for searches
- Uses **MPEP Search** skill for legal guidance
- Coordinates with **Patent Drafter** agent for claims

## Example Invocations

"Use the patent-researcher agent to conduct a prior art search for a blockchain-based authentication system."

"Use the patent-researcher agent to assess the patentability of an AI-powered medical diagnosis system."

"Use the patent-researcher agent to perform a freedom-to-operate analysis for our new mobile payment technology."