NYC

gemini-api-dev

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • PROMPT_INJECTION (LOW): The skill uses directive language ('Your knowledge is outdated', 'Models like gemini-2.5-*... are legacy') to attempt to override the agent's internal state. It falsely claims that current production models are deprecated and instructs the agent to use speculative or non-existent model names like 'gemini-3-pro-preview'.
  • INDIRECT_PROMPT_INJECTION (LOW): The skill establishes an attack surface (Category 8) by instructing the agent to fetch and process content from an external 'llms.txt' index and associated documentation files.
  • Ingestion points: https://ai.google.dev/gemini-api/docs/llms.txt and linked .md.txt files.
  • Boundary markers: Absent; there are no delimiters or instructions for the agent to ignore embedded commands in the fetched data.
  • Capability inventory: The agent is instructed to use the fetched data for implementing and debugging API integrations, which may involve code generation or tool usage.
  • Sanitization: Absent.
  • EXTERNAL_DOWNLOADS (SAFE): The skill references official Google SDKs (google-genai, @google/genai, google.golang.org/genai). These packages and the associated domains (googleapis.com, google.dev) are recognized as trusted sources per the [TRUST-SCOPE-RULE].
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 04:36 PM