deep-init

Extract domain rules from PR comments, tickets, and wiki ADRs into CLAUDE.md.

Updated Feb 19, 2026
One-click install
npx skills add https://github.com/Cronan/claude-code-prompts --skill deep-init
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: deep-init
Source: https://github.com/Cronan/claude-code-prompts/tree/main/skills/deep-init
Command: npx skills add https://github.com/Cronan/claude-code-prompts --skill deep-init

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Claude Code's /init output often misses domain rules learned from real-world post-mortems. /deep-init audits git history, PR comments, issue tracker tickets, and team wiki pages to produce a CLAUDE.md grounded in the team's actual experience.

Core Features & Use Cases

  • Audit sources: mines git history, PRs, tickets, and wiki ADRs to extract domain rules that reflect how the team actually works.
  • Generate rules: synthesizes the findings into concise, prioritized CLAUDE.md instructions plus optional reference materials for architecture, glossary, and edge cases.
  • Safe upgrade path: provides a route to replace an existing CLAUDE.md with a domain-aware version while preserving valuable /init content.

Quick Start

Run the audit workflow to replace your CLAUDE.md with domain-specific rules derived from real PRs, tickets, and wiki content.

Frequently Asked Questions about deep-init

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate a CLAUDE.md file from real project history and PR comments?

You can generate a domain-aware CLAUDE.md by mining git history, PR comments, and closed tickets to extract team rules, then synthesizing those findings into a root configuration file with scoped, codified instructions that link directly to code elements.

What is the best way to capture domain rules from team wiki ADRs and post-mortems for Claude Code?

The best way to capture domain rules from wiki ADRs is to audit them alongside git history and tickets, extracting architectural decisions and edge cases into a concise CLAUDE.md with optional reference materials for architecture, glossary, and edge cases.

Can I upgrade an existing CLAUDE.md without losing valuable /init content?

Yes, you can upgrade an existing CLAUDE.md safely by using a workflow that replaces your current file with a domain-aware version while preserving valuable /init content, ensuring no existing instructions are lost during the refactor.

Does deep-init work with closed tickets and issue tracker history to extract code instructions?

Yes, it works with closed tickets and issue tracker history by mining them alongside PR comments and wiki pages to extract domain rules, synthesizing the findings into prioritized CLAUDE.md instructions with references to specific code elements and files.

Why does my Claude Code /init output miss domain-specific rules learned from post-mortems?

Claude Code /init output misses domain-specific rules because it does not audit historical team sources. To capture them, you need a workflow that mines git history, PRs, tickets, and wiki ADRs to produce a CLAUDE.md grounded in actual team experience.