Uspto DatabaseSAFE

Intelligent Skill routing and workflow orchestration for AI agents — +21.12 pp reward, −29.6% tokens on SkillsBench with DeepSeekV4Flash-VE.

The instruction file, verbatim from the audited commit — this is the text the model reads, and the surface the audit's instruction layer examines. Quoted here so you can judge it without cloning anything.

---
name: uspto-database
description: Access USPTO APIs for patent/trademark searches, examination history (PEDS), assignments, citations, office actions, TSDR, for IP analysis and prior art searches.
license: Unknown
metadata:
    skill-author: K-Dense Inc.
---

# USPTO Database

## Overview

USPTO provides specialized APIs for patent and trademark data. Search patents by keywords/inventors/assignees, retrieve examination history via PEDS, track assignments, analyze citations and office actions, access TSDR for trademarks, for IP analysis and prior art searches.

## When to Use This Skill

This skill should be used when:

- **Patent Search**: Finding patents by keywords, inventors, assignees, classifications, or dates
- **Patent Details**: Retrieving full patent data including claims, abstracts, citations
- **Trademark Search**: Looking up trademarks by serial or registration number
- **Trademark Status**: Checking trademark status, ownership, and prosecution history
- **Examination History**: Accessing patent prosecution data from PEDS (Patent Examination Data System)
- **Office Actions**: Retrieving office action text, citations, and rejections
- **Assignments**: Tracking patent/trademark ownership transfers
- **Citations**: Analyzing patent citations (forward and backward)
- **Litigation**: Accessing patent litigation records
- **Portfolio Analysis**: Analyzing patent/trademark portfolios for companies or inventors

## USPTO API Ecosystem

The USPTO provides multiple specialized APIs for different data needs:

### Core APIs

1. **PatentSearch API** - Modern ElasticSearch-based patent search (replaced legacy PatentsView in May 2025)
   - Search patents by keywords, inventors, assignees, classifications, dates
   - Access to patent data through June 30, 2025
   - 45 requests/minute rate limit
   - **Base URL**: `https://search.patentsview.org/api/v1/`

2. **PEDS (Patent Examination Data System)** - Patent examination history
   - Application status and transaction history from 1981-present
   - Office action dates and examination events
   - Use `uspto-opendata-python` Python library
   - **Replaced**: PAIR Bulk Data (PBD) - decommissioned

3. **TSDR (Trademark Status & Document Retrieval)** - Trademark data
   - Trademark status, ownership, prosecution history
   - Search by serial or registration number
   - **Base URL**: `https://tsdrapi.uspto.gov/ts/cd/`

### Additional APIs

4. **Patent Assignment Search** - Ownership records and transfers
5. **Trademark Assignment Search** - Trademark ownership changes
6. **Enriched Citation API** - Patent citation analysis
7. **Office Action Text Retrieval** - Full text of office actions
8. **Office Action Citations** - Citations from office actions
9. **Office Action Rejection** - Rejection reasons and types
10. **PTAB API** - Patent Trial and Appeal Board proceedings
11. **Patent Litigation Cases** - Federal district court litigation data
12. **Cancer Moonshot Data Set** - Cancer-related patents

## Quick Start

### API Key Registration

USPTO APIs require an API key. Register at:
**https://account.uspto.gov/api-manager/**

API key for **PatentSearch API** is provided by PatentsView. Register at:
**https://patentsview.org/api-v01-information-page**

Set the API key as an environment variable:
```bash
export USPTO_API_KEY="your_api_key_here"
export PATENTSVIEW_API_KEY="you_api_key_here"
```

### Helper Scripts

This skill includes Python scripts for common operations:

- **`scripts/patent_search.py`** - PatentSearch API client for searching patents
- **`scripts/peds_client.py`** - PEDS client for examination history
- **`scripts/trademark_client.py`** - TSDR client for trademark data

## Task 1: Searching Patents

### Using the PatentSearch API

The PatentSearch API uses a JSON query language with various operators for flexible searching.

#### Basic Patent Search Examples

**Search by keywords in abstract:**
```python
from scripts.patent_search import PatentSearchClient

client = PatentSearchClient()

# Search for machine learning patents
results = client.search_patents({
    "_text_all": {"patent_abstract": "machine learning"}
})

for patent in results['patents']:
    print(f"{patent['patent_number']}: {patent['patent_title']}")
```

**Search by inventor:**
```python
results = client.search_by_inventor("John Smith")
```

**Search by assignee/company:**
```python
results = client.search_by_assignee("Google")
```

**Search by date range:**
```python
results = client.search_by_date_range("2024-01-01", "2024-12-31")
```

**Search by CPC classification:**
```python
results = client.search_by_classification("H04N")  # Video/image tech
```

#### Advanced Patent Search

Combine multiple criteria with logical operators:

```python
results = client.advanced_search(
    keywords=["artificial", "intelligence"],
    assignee="Microsoft",
    start_date="2023-01-01",
    end_date="2024-12-31",
    cpc_codes=["G06N", "G06F"]  # AI and computing classifications
)
```

#### Direct API Usage

For complex queries, use the API directly:

```python
import requests

url = "https://search.patentsview.org/api/v1/patent"
headers = {
    "X-Api-Key": "YOUR_API_KEY",
    "Content-Type": "application/json"
}

query = {
    "q": {
        "_and": [
            {"patent_date": {"_gte": "2024-01-01"}},
            {"assignee_organization": {"_text_any": ["Google", "Alphabet"]}},
            {"cpc_subclass_id": ["G06N", "H04N"]}
        ]
    },
    "f": ["patent_number", "patent_title", "patent_date", "inventor_name"],
    "s": [{"patent_date": "desc"}],
    "o": {"per_page": 100, "page": 1}
}

response = requests.post(url, headers=headers, json=query)
results = response.json()
```

### Query Operators

- **Equality**: `{"field": "value"}` or `{"field": {"_eq": "value"}}`
- **Comparison**: `_gt`, `_gte`, `_lt`, `_lte`, `_neq`
- **Text search**: `_text_all`, `_text_any`, `_text_phrase`
- **String matching**: `_begins`, `_contains`
- **Logical**: `_and`, `_or`, `_not`

**Bes

SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.

What the source does

Filesystem

none-observed

Network

declared (3 observation(s))

Shell

none-observed

Dependencies

pinned

Secrets in source

none-found

Findings (0)

No findings outside the package's declared scope.

Gates applied: no_behavioural_pass.