The Agent Skills Directory

[SAFE]: The skill implements a robust 'Research-Decide-Plan-Implement' workflow for managing multiple Git repositories. It includes strict requirements for mechanical verification (diffs, hashes) of all claims and mandates user approval for all changes before implementation. Security-relevant features like Git operations and parallelization are scoped correctly to their intended purpose. The use of $codex-handoff and $commit dependencies suggests integration with standard, managed agent tools. Pre-existing file changes are explicitly protected from modification, and commit operations are semantic and scoped to specific task files. No obfuscation, prompt injection, or unauthorized network activity was found.

repo-harmonization