The Agent Skills Directory

[SAFE]: The skill provides instructions for using a structured repo-analyze tool to clone and examine GitHub repositories. It encourages targeted analysis rather than recursive operations on unknown content.
[COMMAND_EXECUTION]: The skill contains various shell commands (grep, find, cat, git, ls, rm) for analyzing repository structures, dependencies, and patterns. These are standard development tools and are used within a controlled workflow (e.g., targetting /tmp/repo).
[EXTERNAL_DOWNLOADS]: The skill facilitates the cloning of GitHub repositories for analysis purposes. It specifically targets the GitHub domain and uses the official gh CLI or a dedicated internal tool, which is an expected behavior for a repository analysis skill.
[REMOTE_CODE_EXECUTION]: While the skill analyzes remote code, it does not instruct the agent to execute the code it downloads. The workflow focuses on static analysis, documentation review, and metadata extraction (grep, cat, find).
[DATA_EXFILTRATION]: No patterns of exfiltration were detected. The network operations are limited to cloning public repositories from GitHub for analysis.
[CLEANUP_PRACTICES]: The skill explicitly instructs the user to clean up temporary directories (rm -rf /tmp/repo-name) to prevent disk accumulation, which is a good security and operational practice.

repo-analysis