Review the given Gitlab MR (provided after this, use Zapier MCP to review the changes).

When to Use This Skill

Activate this skill when:

The user types "review" or "code review" (with or without slash command)
The user types "review BRANCH-NAME" to review a specific branch
The user types "review TICKET-ID" (e.g., "review AGP-123" or "review AICC-456") to review the branch associated with a Jira ticket
The user types "review LINK_TO_GITLAB_MR", in this case use Zaper MCP to fetch the MR details
The user asks to review a branch, pull request, or merge request

Keep in mind I suspect this code is 100-90% AI generated (with some light human in the loop) and as such look out for halmarks of AI generated code. They mostly boil down to LLM's optimizing for token generation (it's cheap) whereas humas optimize for readability and maintainability (less tokens often leads to this as reading andwriting tokens is far more expensive for a human than an LLM).

1. Review for LLM optimizing for tokens rather than humans

Example: MR contains changes to a types.ts file, the LLM re-used a list of strings 4 times instead of creating an enum type. That's because the repetition was cheap for the LLM, but a human would never do that because we'd realize that if we ever needed to add something to this list we'd need to add it in 4+ places and for a human that takes time.

Token-inefficient patterns:

Repeating code blocks instead of extracting functions/types after 2+ uses
Verbose names and excessive comments restating obvious code
Try-catch blocks everywhere, redundant null checks, defensive code that adds no value
Over-explicit types that could be inferred

code-review

When to Use This Skill

1. Review for LLM optimizing for tokens rather than humans