What are the technical requirements to run this skill?

You must have the GitHub CLI (gh) authenticated on your machine and Node.js version 18 or higher installed.

Why does it require 3+ occurrences of a pattern?

The three-occurrence threshold ensures that the patterns identified are established team conventions rather than one-off feedback or situational nitpicks.

Can I analyze PRs from a specific developer?

Yes, the skill supports an 'author' parameter, allowing you to focus on patterns established by specific senior developers or architects.

Does it automatically overwrite my project files?

No, it generates a proposal of AGENTS.md rules with citations. You maintain control over which patterns are ultimately committed to your repository's ruleset.

How does this skill differentiate between bot and human comments?

The skill utilizes specific filtering logic to exclude known bot accounts and automated messages, ensuring the resulting patterns reflect genuine human reviewer intent.

Learn From PRs

Name: Learn From PRs
Author: vltansky

byvltansky

协作工具

Analyzes pull request review history to extract team-specific coding patterns and automatically generate documentation for AI agents.

关于

Learn From PRs is a powerful utility designed to bridge the gap between human expertise and AI behavior by mining GitHub pull request history for recurring feedback. It intelligently filters out bot noise and generic process comments to focus on specific architectural, API, and styling preferences unique to your codebase. By converting these patterns into actionable rules for AGENTS.md with verifiable citations, it ensures that Claude and other AI tools adhere to your team's established conventions from day one.

主要功能

Direct generation of AGENTS.md formatted rules with clickable PR citations
Semantic pattern recognition that groups similar reviewer feedback into cohesive rules
0 GitHub stars
Support for author-specific or repository-wide analysis using the GitHub CLI
Intelligent filtering to exclude bot noise and non-code process feedback
Auto-escalation logic that analyzes up to 1000 PRs to identify statistically significant patterns

使用场景

Identifying recurring technical debt or API misuse patterns highlighted in code reviews
Onboarding new team members by documenting unwritten code conventions and reviewer preferences
Setting up AGENTS.md for existing projects to ensure AI-generated code matches team standards

关于

主要功能

Direct generation of AGENTS.md formatted rules with clickable PR citations
Semantic pattern recognition that groups similar reviewer feedback into cohesive rules
0 GitHub stars
Support for author-specific or repository-wide analysis using the GitHub CLI
Intelligent filtering to exclude bot noise and non-code process feedback
Auto-escalation logic that analyzes up to 1000 PRs to identify statistically significant patterns

使用场景

Identifying recurring technical debt or API misuse patterns highlighted in code reviews
Onboarding new team members by documenting unwritten code conventions and reviewer preferences
Setting up AGENTS.md for existing projects to ensure AI-generated code matches team standards