The Agent Skills Directory

[SAFE]: No malicious patterns, obfuscation, or data exfiltration attempts were detected. The skill's behavior matches its stated purpose of managing reference repositories.
[COMMAND_EXECUTION]: The skill executes standard CLI tools such as git, gh, grep, find, and ls via helper scripts (clone-or-update.sh and sync-all.sh). These operations are restricted to the ~/projects/referenzen/ directory and are used for repository management and read-only searches.
[EXTERNAL_DOWNLOADS]: The skill fetches repository data from GitHub using the official gh and git tools. These are well-known, trusted services, and the operations use the user's existing authentication.
[PROMPT_INJECTION]: As the skill is designed to read and process content from external Git repositories, it presents a surface for Indirect Prompt Injection. Maliciously crafted content within a referenced repository could theoretically influence agent behavior. The skill includes explicit instructions to avoid exposing secrets like .env files, which mitigates data exposure risks.

project-references