The Agent Skills Directory

[SAFE]: No malicious patterns or security vulnerabilities were identified in the skill instructions or logic.
[DATA_EXFILTRATION]: The skill reads source code changes via git diff to generate commit messages. This is the intended behavior and the data remains within the local environment without being transmitted externally.
[PROMPT_INJECTION]: The skill processes untrusted data from git diffs which could potentially contain adversarial instructions (Indirect Prompt Injection). However, the skill's restricted capabilities prevent any harmful actions from being executed, as it can only output text and lacks tools to modify files or access the network.

commit-message