The Agent Skills Directory

[DATA_EXPOSURE]: The skill instructs the agent to check the existence of local configuration files (e.g., ~/.gemini/settings.json) and the status of development tools (e.g., gh auth status, npm --version) to determine available tools. These are benign environmental checks used to scope the agent's capabilities.
[EXTERNAL_DOWNLOADS]: The workflow encourages searching for and suggesting the adoption of third-party libraries from official package registries like npm and PyPI. This is a standard development practice and the skill references well-known, reputable packages as examples.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to search external, potentially untrusted sources like package registries and the web. This creates a surface where external content enters the agent context. This is an inherent part of the research functionality and does not represent a malicious bypass or autonomous execution risk.

search-first