alterlab-uspto
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill interacts exclusively with official USPTO and PatentsView API endpoints (uspto.gov and patentsview.org) to retrieve intellectual property data.
- [SAFE]: Security best practices for secret management are followed by requiring users to set API keys as environment variables (USPTO_API_KEY, PATENTSVIEW_API_KEY) rather than hardcoding credentials in scripts.
- [SAFE]: External dependencies are restricted to the standard requests library and the specialized, verifiable uspto-opendata-python package for processing examination records.
- [INDIRECT_PROMPT_INJECTION]: The skill provides a surface for indirect prompt injection as it ingests untrusted third-party content (patent abstracts, claims, trademark descriptions) from USPTO API responses.
- Ingestion points: API response parsing in scripts/patent_search.py, scripts/trademark_client.py, and scripts/peds_client.py.
- Boundary markers: Absent; external data is returned directly into the agent context.
- Capability inventory: Network access via requests and local Python script execution.
- Sanitization: None; the skill retrieves and displays raw text fields from the database.
- [SAFE]: No obfuscation, persistence mechanisms, or unauthorized privilege escalation patterns were detected in the skill instructions or scripts.
Audit Metadata