The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local shell script (smart_commit.sh) to wrap Git commands including git add, git commit, and git push.
[DATA_EXFILTRATION]: The skill performs git push operations to transmit local repository content to remote servers. This is the intended primary purpose of the skill and targets standard Git remotes (e.g., GitHub).
[DATA_EXPOSURE]: The script uses git add . to stage all changes in the current directory. This command can inadvertently stage and push sensitive files such as credentials, .env files, or private keys if they are not properly excluded in the repository's .gitignore file.
[PROMPT_INJECTION]: The skill facilitates an indirect prompt injection surface by interpolating user-provided commit messages into shell commands.
Ingestion points: User-provided commit message string passed as an argument to smart_commit.sh (as seen in SKILL.md).
Boundary markers: Absent. The message is passed as a positional argument without delimiters or instructions to the agent to ignore embedded commands.
Capability inventory: The skill can execute shell scripts and perform network operations via git push.
Sanitization: None. The commit message is used directly in the git commit -m "$MESSAGE" command.

git-pushing