Review patterns
npx -y skills add mrge-io/skills --skill review-patternsAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
- 2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Surfaces the team's coding patterns and conventions learned by cubic from senior reviewer feedback and codebase analysis. Use when the user is writing code, asking about team standards, or wondering "how does this team do X".
SKILL.md
2.3 KB, 424 tokens by cl100k_base, as published. Nobody here has run it
Team Review Patterns
Surface the team's coding conventions and review preferences from cubic's learned patterns.
When To Activate
- User is writing new code and wants it to match team standards
- User asks about team coding conventions or review preferences
- User is preparing code for review and wants to preemptively fix issues
- User wonders why cubic flags certain patterns
Inputs
- Context (optional): What the user is currently working on, to filter relevant learnings.
Instructions
1. Detect the repository
git remote get-url origin
Extract the owner and repo name.
2. Fetch learnings
Call list_learnings with the owner and repo.
3. Filter by relevance
Match learning categories to the user's current task:
- Prioritize high-confidence learnings
- Focus on learnings from senior reviewer analysis (highest signal)
- If the user is working on a specific file or module, prioritize related learnings
4. Enrich if needed
If a specific learning is highly relevant, call get_learning for full details including the original review feedback that generated it.
Output Format
Present learnings as team preferences, not rigid rules. Group related learnings and include confidence levels.
## Team Patterns for this repo
### Error Handling (High Confidence)
- **Always wrap external API calls in try/catch** — The team prefers explicit error
handling over global middleware. Learned from senior reviewer feedback.
- **Use custom error classes** — Throw `AppError` subclasses, not raw `Error`.
### Naming (Medium Confidence)
- **Boolean variables use `is`/`has` prefix** — e.g., `isActive`, `hasPermission`.
### Testing (High Confidence)
- **Integration tests for all API endpoints** — Unit tests alone are not sufficient
for route handlers.
_3 learnings from senior reviewer analysis · 1 from user feedback_
If no learnings exist, explain that cubic learns patterns over time from senior reviewer feedback.