The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The documentation references installation of well-known machine learning libraries such as TensorBoard, PyTorch, and TensorFlow from official package registries.
[INDIRECT_PROMPT_INJECTION]: The skill provides interfaces for logging text and metadata from model training processes. This creates an ingestion surface for potentially untrusted data that could be displayed in the TensorBoard UI.
Ingestion points: Data logging functions like writer.add_text and writer.add_embedding in SKILL.md and references/visualization.md.
Boundary markers: No specific delimiters or warnings for embedded instructions are included in the boilerplate examples.
Capability inventory: The skill facilitates writing training logs to the local file system for visualization.
Sanitization: The provided code examples do not perform sanitization on text inputs before logging.
[REMOTE_CODE_EXECUTION]: The skill includes instructions for running the standard tensorboard command-line utility to visualize log directories.

tensorboard