The Agent Skills Directory

[COMMAND_EXECUTION] (HIGH): The skill defines autoExecute: true and scriptPath: check-git.sh in its YAML frontmatter. Because the actual script content is not provided for auditing, this configuration represents an unverified execution path that runs automatically upon Git-related triggers.
[PROMPT_INJECTION] (HIGH): This skill is vulnerable to Indirect Prompt Injection (Category 8).
Ingestion points: Triggers on untrusted data from git commit and commit message inputs.
Boundary markers: None present; the skill treats commit messages as direct input.
Capability inventory: Has the capability to execute Bash scripts (scriptType: bash).
Sanitization: No evidence of sanitization or escaping for the ingested commit data before it interacts with the execution environment.
[EXTERNAL_DOWNLOADS] (MEDIUM): The documentation instructs users to execute npm install -g commitizen cz-conventional-changelog. These are unversioned global installations of third-party packages, which increases the supply chain risk and introduces external code into the user's environment.

git-commit-guide