The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses gh api and git commands to configure repository settings and detect branch names. These are appropriate for the skill's stated purpose of repository hardening and follow the principle of least privilege by checking for existing configurations before applying new ones.
[REMOTE_CODE_EXECUTION]: There is no remote code execution detected. All scripts are provided locally within the skill's scripts/ directory and are copied into the user's repository for local execution.
[DATA_EXFILTRATION]: No data exfiltration patterns were found. Network activity is limited to the gh api calls to GitHub's official API for repository configuration, which is a trusted and expected operation.
[OBFUSCATION]: The scripts and instructions are written in clear, plain text with no evidence of obfuscation, hidden characters, or encoded payloads.
[PROMPT_INJECTION]: The instructions are clear and do not attempt to override the AI's safety guidelines or core personality. It uses natural instructional language to guide the setup process.
[PRIVILEGE_ESCALATION]: The skill does not request elevated permissions (like sudo). It uses standard development tools (gh, git, node) within the user's project environment.
[INDIRECT_PROMPT_INJECTION]: The skill includes a hook surface in .claude/settings.json, but it is configured to execute specific, local, static scripts (block-destructive-git.sh) provided by the skill author rather than arbitrary data from files, effectively mitigating injection risks.

setup-repo