The Agent Skills Directory

[SAFE]: The skill implements a robust safety framework for git operations. It explicitly categorizes commands into safe, caution, and forbidden groups, strictly prohibiting destructive actions such as git reset --hard, git push, or git clean -f to protect the integrity of the user's repository.\n- [COMMAND_EXECUTION]: The skill utilizes git bisect run to automate regression detection, which involves executing shell-based test commands. To mitigate risks, the instructions mandate that the agent must validate test commands, perform dry-runs, and obtain explicit user confirmation before execution.\n- [SAFE]: The skill includes instructions to detect and warn the user about potential credential exposure within git history, aligning with security best practices for analyzing legacy or AI-assisted commits.

rewind