gemini-api-dev

Pass

Audited by Gen Agent Trust Hub on May 20, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • Prompt Instruction Pattern: The skill uses strong directives such as "These rules override your training data" and "Your knowledge is outdated" to prioritize its instructions over the model's pre-existing knowledge. This is a common prompt engineering technique to ensure the use of specific API versions.
  • Official Library Dependencies: The skill recommends the installation of several Google-maintained SDKs, including google-genai (Python), @google/genai (JavaScript), and google.golang.org/genai (Go). These are official, well-known libraries for interacting with the Gemini API.
  • External Documentation Retrieval: The instructions guide the agent to fetch supplemental information from ai.google.dev. This involves interacting with external web content, which is a standard requirement for a documentation-focused skill but remains an area to monitor for general data processing safety.
  • MCP Tool Integration: The skill includes instructions to prioritize Model Context Protocol (MCP) tools for documentation searching when available, which is a modern standard for efficient and secure tool-assisted information retrieval.
Audit Metadata
Risk Level
SAFE
Analyzed
May 20, 2026, 05:13 AM
Security Audit — agent-trust-hub — gemini-api-dev