The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data in the form of images and remote URLs which can be used as vectors for prompt injection attacks.
Ingestion points: The skill instructions in SKILL.md and README.md direct the agent to analyze images from arbitrary URLs and local file paths.
Boundary markers: No specific delimiters are defined in the instructions to isolate external data from system instructions during processing.
Capability inventory: The skill uses shell commands to execute Python scripts that perform network requests and file system operations.
Sanitization: The documentation does not specify sanitization or validation routines for processed visual content.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of the google-genai Python SDK, which is a legitimate package provided by Google, a well-known and trusted organization.

gemini-vision