The Agent Skills Directory

[SAFE]: The skill operates locally within a software repository, using standard Git commands to gather commit data. No unauthorized data access or external communication was detected.
[COMMAND_EXECUTION]: The skill utilizes git log to retrieve history based on user-defined time ranges. The instructions specify that the agent should format these inputs into strict date strings (YYYY-MM-DD) or ISO 8601 timestamps, which aligns with safe command execution practices.
[DATA_EXFILTRATION]: No network activity was detected. All data processing is local, and the resulting changelog is stored in a temporary local directory (tmp/).
[PROMPT_INJECTION]: The skill processes commit messages, which are external data. While commit messages could contain malicious text, the skill's functionality is limited to summarization into a static markdown file, and it does not provide any capability for the content of those messages to be executed as instructions.

superplane-changelog