The Agent Skills Directory

[Indirect Prompt Injection] (HIGH): The skill is highly vulnerable to malicious instructions embedded in git commit messages. * Ingestion points: Git commit history (untrusted external source) and local configuration files like CHANGELOG_STYLE.md. * Boundary markers: None present to isolate untrusted commit text from agent instructions. * Capability inventory: Requires filesystem read access and potentially write access if used to update CHANGELOG.md directly. * Sanitization: No validation or filtering of commit content is performed.
[Command Execution] (LOW): The skill requires the ability to execute git commands to retrieve repository history, which is a standard but sensitive capability.

changelog-generator