review-patterns
Scannednpx machina-cli add skill mrge-io/cubic-claude-plugin/review-patterns --openclawTeam Review Patterns
This skill pulls team-specific code review learnings from cubic to apply the team's conventions when writing or reviewing code.
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
How to Use
- Detect the current repository from git remote:
git remote get-url origin - Call
list_learningswith the owner and repo - Filter learnings by relevance to the user's current task:
- Match learning categories to the type of code being written
- Prioritize high-confidence learnings
- Focus on learnings from senior reviewer analysis (highest signal)
- If a specific learning is relevant, call
get_learningfor full details including the original feedback
Presentation
- Present learnings as team preferences, not rigid rules
- Connect each learning to the user's current code when possible
- Group related learnings together
- Mention the confidence level so the user can judge how strongly the team feels about each pattern
- If no learnings exist, explain that cubic learns patterns over time from senior reviewer feedback
Source
git clone https://github.com/mrge-io/cubic-claude-plugin/blob/main/skills/review-patterns/SKILL.mdView on GitHub Overview
This skill pulls team-specific code review learnings from Cubic to apply the team's conventions when writing or reviewing code. It helps developers align with standards, preempt issues before submission, and understand what patterns the team cares about in reviews. By surfacing confidence levels, it guides you to the most impactful items.
How This Skill Works
It detects the current repository via git remote, then calls list_learnings with the owner and repo. It filters learnings by relevance to the user's task, prioritizes high-confidence learnings and senior reviewer analysis, and calls get_learning for full details when a match is found.
When to Use It
- User is writing new code to match team standards
- User asks about team coding conventions or review preferences
- User is preparing code for review to preemptively fix issues
- User wonders why Cubic flags certain patterns
- New to the project and onboarding to Cubic-based review patterns
Quick Start
- Step 1: Detect the current repository from git remote: git remote get-url origin
- Step 2: Call list_learnings with the owner and repo
- Step 3: Filter learnings by relevance to the current task and call get_learning for full details if a match
Best Practices
- Present learnings as team preferences, not rigid rules
- Connect each learning to the user's current code when possible
- Group related learnings together for easier scanning
- Mention the confidence level for each learning so the user can gauge importance
- Prioritize high-confidence learnings and those from senior reviewer analysis
Example Use Cases
- A developer adds a new module; Cubic returns learnings to enforce naming and simplicity rules and flags potential anti-patterns.
- A reviewer asks Cubic for patterns the team cares about in code quality, to prepare constructive feedback.
- Before submitting a PR, a user applies learnings to preemptively fix formatting and edge-case handling.
- A team lead cites senior reviewer learnings to justify why a pattern is preferred in reviews.
- A newcomer uses review-patterns to quickly understand the team's conventions and review expectations.
