Can I analyze only a specific part of my project?

Yes, the skill supports a --scope flag to focus analysis on specific directories (like /api or /testing) and a --since flag for date-based filtering.

How many commits are needed for accurate rule derivation?

A minimum of 20 commits is recommended for the skill to identify meaningful patterns, though accuracy increases with the depth of the history.

Does it require a specific commit message format?

While it works with any history, it performs significantly better with projects using Conventional Commits, as it can specifically target 'feat!', 'fix', and 'refactor' types.

How does this skill handle conflicting decisions in git history?

The skill applies a 'latest wins' strategy where decisions extracted from more recent commits automatically override older patterns to reflect the current state of the project.

Where are the derived rules stored?

Rules are generated as structured Markdown files within the .claude/rules/ directory, making them immediately accessible to Claude for future tasks.

Blueprint: Derive Project Rules

Name: Blueprint: Derive Project Rules
Author: laurigates

bylaurigates

•

Developer Tools

Automates the extraction of project coding standards and architectural decisions from git commit history to establish consistent AI development guidelines.

The blueprint-derive-rules skill enables Claude to perform deep forensic analysis of your project's git history to uncover implicit development patterns and architectural decisions. By scanning commit messages and diffs—specifically targeting refactors, fixes, and breaking changes—it codifies 'tribal knowledge' into structured Claude rules stored in your repository. The skill intelligently handles project evolution by ensuring newer commits override older decisions, resulting in an AI assistant that is perfectly aligned with your team's current coding standards and best practices.

Key Features

0110 GitHub stars

02Smart conflict resolution where recent commits supersede legacy patterns

03Automated decision mining from git commit logs and diffs

04Domain-specific rule categorization for testing, API design, and code style

05Commit quality reporting and conventional commit percentage tracking

06Parallel agent analysis for high-efficiency history scanning

Use Cases

01Automatically updating AI behavior following major project refactors or breaking changes

02Standardizing development guidelines across teams by formalizing historical decisions

03Onboarding Claude to a legacy codebase with undocumented architectural patterns

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add laurigates/claude-plugins blueprint-derive-rules

For use in Claude.ai and ChatGPT

Download Skill