alphaear-search

Pass

Audited by Gen Agent Trust Hub on May 7, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill uses established search engines (DuckDuckGo, Baidu) and the Jina Search/Reader APIs for information retrieval. These interactions are standard for the skill's purpose and do not involve unauthorized data exfiltration.
  • [SAFE]: Sensitive configuration, such as API keys for LLM providers (OpenAI, DeepSeek, Zai, etc.) and the Jina service, are managed strictly via environment variables, following secure coding practices.
  • [SAFE]: Local data operations are restricted to a SQLite database used for search result caching and RAG functionality, with no access to sensitive system files or credentials.
  • [SAFE]: External model loading for sentiment analysis and embeddings is performed through the standard Transformers library from trusted HuggingFace repositories.
  • [SAFE]: The skill's architecture uses modular scripts for LLM routing and capability testing, ensuring that tool calls and reasoning tasks are directed to appropriate models securely.
Audit Metadata
Risk Level
SAFE
Analyzed
May 7, 2026, 05:57 PM