How does the skill detect new rules during a conversation?

The skill monitors for specific linguistic triggers such as 'always do X', 'the standard is Y', and even frustration signals like 'stop doing that' to identify when a user is establishing a new preference.

How do I review and approve the rules the AI has captured?

You can use the 'clean up rules' or 'process captured rules' command to trigger a review session where you can approve, modify, or delete pending rules.

Where are the captured rules stored?

Rules are initially saved to a pending file at .agents/SYSTEM/CAPTURED-RULES.md and can be promoted to permanent files like USER-PREFERENCES.md or RULES.md after user confirmation.

Does this skill automatically update my source code?

No, this skill is focused on documentation and behavior guidance. It captures rules to ensure future AI responses align with your standards, rather than performing automated refactoring.

Rules Capture

Name: Rules Capture
Author: shipshitdev

byshipshitdev

•

Productivity & Workflow

Automatically identifies and documents coding preferences, style guidelines, and project rules from natural conversation.

The Rules Capture skill acts as an intelligent memory layer for your development workflow by monitoring conversations for explicit instructions, preferred patterns, and frustration signals. It parses these interactions to extract structured rules—categorizing them by coding style, workflow, tools, or communication—and stages them for your review before promoting them to permanent project documentation. This ensures that your AI assistant learns your specific standards over time, reducing repetitive corrections and maintaining consistency across your codebase without manual documentation overhead.

Key Features

01Integrated cleanup commands to bulk process and promote captured rules

0210 GitHub stars

03Two-stage verification system with a pending review queue and permanent storage

04Structured data extraction including category, rule type, context, and examples

05Sentiment-aware capture of frustration signals to identify and correct repetitive errors

06Real-time detection of direct rule expressions and coding patterns

Use Cases

01Capturing critical 'Never Do' rules to prevent the AI from repeating known anti-patterns

02Building a personalized user preferences profile based on iterative feedback during development

03Standardizing team-specific architectural patterns and naming conventions across a repository

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add shipshitdev/library rules-capture

For use in Claude.ai and ChatGPT

Download Skill