The Agent Skills Directory

[SAFE]: The skill integrates exclusively with the official Weights & Biases platform using established SDKs (wandb and weave).
[SAFE]: All network activity is directed toward legitimate Weights & Biases API endpoints for artifact management and run logging.
[SAFE]: Data processing logic includes explicit safety boundaries, such as a 10,000-row and 100-column cap, to prevent excessive resource consumption or agent instability.
[SAFE]: The skill uses standard environment variables (WANDB_API_KEY, WANDB_ENTITY, and WANDB_PROJECT) for authentication, following secure configuration practices.
[SAFE]: No patterns of obfuscation, unauthorized persistence, or privilege escalation were found in the instructional content or the implementation scripts.

wandb-eval-tables