The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external sources, such as model archives and datasets, which could contain adversarial instructions designed to influence the agent's behavior.
Ingestion points: Dataset retrieval and ingest functions (client.datasets.retrieve and client.datasets.ingest in references/recipes.md) which process data from local archives or remote URLs.
Boundary markers: The instructions do not specify the use of delimiters or clear directives to ignore instructions embedded within the processed data.
Capability inventory: The skill has capabilities for network operations (httpx.put), cloud resource creation (client.training.start, client.exports.create), and model deployment.
Sanitization: There is no explicit evidence of content sanitization or validation for datasets retrieved from the platform or remote sources.
[EXTERNAL_DOWNLOADS]: The skill instructions include installing the ultralytics-platform and ultralytics Python packages. These are official packages used to interact with the Ultralytics ecosystem and platform services.
[COMMAND_EXECUTION]: Provides instructions for using the yolo CLI for authentication (yolo login) and model training (yolo train). These are standard operational commands for the supported library.

ultralytics-platform