The Agent Skills Directory

[SAFE]: The skill includes a dedicated 'Smart Stage' step that explicitly blacklists sensitive file patterns, including .env, *.pem, *.key, and credentials*, reducing the risk of accidental data exposure.
[COMMAND_EXECUTION]: The skill uses the Bash tool to execute standard Git commands for repository management, such as git status, git diff, git rebase, and git push. All operations are performed on the local repository or the user's defined origin remote.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data by reading the output of git diff to generate commit messages and branch slugs.
Ingestion points: Reads local file changes via git diff and git diff --cached in Step 2 and Step 5 (SKILL.md).
Boundary markers: No explicit delimiters are used to wrap the diff content when processed by the agent.
Capability inventory: The skill has the ability to create branches, commit code, and push to a remote repository via the Bash tool (SKILL.md).
Sanitization: The skill utilizes a shell HEREDOC pattern when executing git commit to prevent content within the generated message from being interpreted as shell commands.

commit