Potential fix for code scanning alert no. 22: Uncontrolled command line

[ArshVermaGit]

Potential fix for https://github.com/ArshVermaGit/SentinelOps-Autonomous-DevOps-AI/security/code-scanning/22

General fix approach: remove dynamic concatenation of command arguments at the process-execution sink and execute only predefined command forms via an internal allowlist/dispatcher. Keep user input limited to validated data fields (here, commit message) inserted only into one explicitly controlled command template.

Best targeted fix in sentinelops-backend/app/services/local_git_service.py:

• Update _run_git so it no longer executes ["git", "-C", repo_path_str] + args.
• Replace it with explicit branching for each supported command:
  • status --short
  • diff --cached --name-only
  • diff --name-only
  • add .
  • push
  • commit -m <sanitized_message>
• Build cmd as a fixed list literal in each branch.
• Keep existing validations (_validate_repo_path_for_fs_access, _ALLOWED_GIT_ARGS, commit sanitizer checks) intact for behavior compatibility and defense in depth.
• Execute subprocess.run(cmd, ...) with that fixed cmd variable.

No changes are required in local_dev.py for this specific sink hardening.

Suggested fixes powered by Copilot Autofix. Review carefully before merging.

[ArshVermaGit]

This looks like a strong sink-hardening improvement with minimal behavioral risk. Replacing dynamic argument concatenation with explicit branching over a small set of supported git command templates makes the execution path much easier to audit and significantly reduces the chance of unintended argument construction. I also like that user input stays confined to the single controlled commit -m <sanitized_message> template while existing validations remain in place for defense in depth. Keeping the command surface explicit (status, diff, add, commit, push) should improve both security posture and maintainability without changing how the service behaves for valid operations.