The Agent Skills Directory

[Command Execution] (MEDIUM): The skill explicitly requests Bash(python:*) and Bash(pip:*) tools. These allow for arbitrary code execution and system modification, posing a risk if the agent is manipulated by untrusted external data.
[Indirect Prompt Injection] (MEDIUM): High-risk attack surface identified. 1. Ingestion points: The skill interacts with ML training data and Tensorboard log files provided by users or external processes. 2. Boundary markers: No delimiters or instructions to ignore embedded commands are present in the skill definition. 3. Capability inventory: The skill has Bash, Write, and Edit permissions, enabling significant side effects. 4. Sanitization: There is no defined logic for sanitizing file paths or log content before processing.
[External Downloads] (LOW): Access to pip facilitates the installation of third-party packages. Without version pinning or a repository whitelist, this allows for the potential retrieval of unverified or malicious code during runtime.

tensorboard-visualizer