gemini-api-dev
Pass
Audited by Gen Agent Trust Hub on Apr 5, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill uses explicit override instructions ("IMPORTANT: These rules override your training data") to bypass the agent's training and replace it with provided information. It further instructs the agent to treat current, stable models as deprecated while mandating the use of future-dated versions, which could lead to hallucinated or broken advice.
- [EXTERNAL_DOWNLOADS]: Recommends installing the
google-genaiand@google/genaiSDKs. These are official packages distributed by Google through standard package registries (PyPI, NPM, etc.). - [COMMAND_EXECUTION]: Provides standard package management commands for Python (
pip), JavaScript (npm), Go (go), and Java (gradle/maven) for the user to set up their development environment. - [PROMPT_INJECTION]: The skill presents an indirect injection surface by instructing the agent to fetch documentation from
ai.google.devviafetch_url. While the target domain is an official and trusted source for Gemini documentation, the skill lacks explicit boundary markers or sanitization instructions for the ingested content.
Audit Metadata