The Agent Skills Directory

[SAFE]: The skill implements strong data privacy guidelines, explicitly forbidding the inclusion of secrets, tokens, or PII (Personally Identifiable Information) in notebooks, logs, or generated artifacts.
[SAFE]: The provided utility script scripts/new.sh demonstrates secure coding practices, including robust input sanitization via sed and tr, and the use of base-10 arithmetic (10#$prefix) to avoid common shell evaluation errors with leading zeros.
[SAFE]: The skill promotes environment isolation by mandating the use of uv for dependency management and environment execution, reducing the risk of global system pollution or dependency conflicts.
[SAFE]: The adapt and verify workflows include mandatory auditing steps, requiring the agent to inspect notebook content for hidden state, unsafe magics, and side effects before execution or refactoring.
[SAFE]: The script generation logic in scripts/new.sh uses deterministic naming and checks for file existence ([[ ! -e "$file" ]]) to prevent accidental overwriting of existing work.

python-notebook-dev