The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes shell commands to perform repository maintenance, specifically invoking git clone, git status, git diff, and git push to synchronize local changes with a remote repository.
[DATA_EXFILTRATION]: The workflow involves copying files from a user-specified source directory to a remote index repository (typically L-yifan/skills). This presents a risk of sensitive data exposure if local directories containing credentials, .env files, or private keys are inadvertently targeted for backup.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection in 'Step 2A', where it reads third-party SKILL.md files and installation instructions from external URLs or marketplaces. These external files could contain instructions designed to manipulate the agent during the synchronization process.
Ingestion points: Upstream SKILL.md files and official installation guides from external GitHub URLs or marketplaces (SKILL.md).
Boundary markers: Absent; the instructions do not implement specific delimiters or 'ignore' directives for processed external content.
Capability inventory: File system read/write operations, directory synchronization, and network-enabled Git operations like git clone and git push (SKILL.md).
Sanitization: Absent; the workflow extracts and records instructions from source files without validation or escaping.

skill-backup