feat: improve code implementaion

.claude/CLAUDE.md

@@ -0,0 +1,170 @@
+# CLAUDE.md
+
+This file provides guidance to Claude Code when working with code in this repository.
+
+## Project Overview
+
+Zerv is a dynamic versioning CLI tool written in Rust that generates versions for any commit from git and other version control systems. It supports multiple version formats (SemVer, PEP440, CalVer) and is designed for CI/CD builds.
+
+---
+
+## 🎯 Critical Rules (Quick Reference)
+
+**MANDATORY - These rules must ALWAYS be followed:**
+
+1. ✅ **Use constants** instead of bare strings (fields, formats, sources, schemas) → Use `crate::utils::constants::*`
+    - **For environment variables**: Use `crate::config::EnvVars::*` (e.g., `EnvVars::PAGER`, `EnvVars::RUST_LOG`)
+2. ✅ **Use `ZervError`** for custom errors and `io::Error::other()` for IO errors
+3. ✅ **Use `get_git_impl()`** for environment-aware Git operations
+4. ✅ **Use `should_run_docker_tests()`** for Docker-dependent tests
+5. ✅ **Never reuse Git implementations** across different directories
+6. ✅ **Include detailed error context** in all error messages
+7. ✅ **NO useless comments** - only comment when code cannot explain itself
+8. ✅ **Check existing utilities** in `src/test_utils/` before creating new ones
+9. ✅ **Place `use` statements at top of file/module** - never inside functions
+10. ✅ **Use `mod` blocks for test organization** - not comment dividers
+
+**NEVER do these:**
+
+1. ❌ Use bare string literals for field/format/source names
+2. ❌ Use `unwrap()` or `expect()` in production code
+3. ❌ Add comments that just repeat function names or restate code
+4. ❌ Create Git fixtures without proper isolation
+5. ❌ Skip Docker test gating for Docker-dependent tests
+6. ❌ Write generic error messages without context
+7. ❌ Duplicate code instead of using existing utilities
+8. ❌ Place `use` statements inside functions (except rare naming conflict cases)
+9. ❌ Use comment dividers for test grouping (use `mod` blocks instead)
+
+**📋 Run `/audit` to check and fix violations automatically**
+
+---
+
+## 📚 Planning Workflow
+
+### Check Existing Plans First
+
+**BEFORE performing ANY coding task, check `.claude/plan/` for existing plans and context.**
+
+Plans in `.claude/plan/` are temporary working documents with short lifecycle:
+
+- Created when planning new features/refactors
+- Used during implementation
+- Deleted or archived when task is complete
+
+### Creating New Plans
+
+**CRITICAL: When user asks you to "write a plan" or "create a plan", you MUST:**
+
+1. ✅ **Check existing `.claude/plan/` files** to find the highest number
+2. ✅ **Create new plan** as `.claude/plan/XX-descriptive-name.md` (increment XX)
+3. ✅ **Do NOT start coding** until the plan is reviewed and approved
+4. ✅ **Use ExitPlanMode tool** to present the plan for approval
+
+**Plan Document Structure:**
+
+- **Status**: Planned/In Progress/Completed
+- **Priority**: High/Medium/Low
+- **Context**: Why this work is needed
+- **Goals**: What we want to achieve
+- **Implementation Plan**: Detailed steps
+- **Testing Strategy**: How to validate
+- **Success Criteria**: Definition of done
+- **Documentation Updates**: What docs need updating
+
+**Note**: Plans can reference `.claude/ref/` docs, but `.claude/ref/` should NEVER reference specific plans (one-way dependency).
+
+---
+
+## 📖 Essential Commands & Workflows
+
+@.claude/ref/workflows/commands.md
+@.claude/ref/workflows/coverage.md
+
+---
+
+## 🏗️ Architecture
+
+### Pipeline Overview
+
+```
+Input → VCS Detection → Version Parsing → Transformation → Format Output
+```
+
+**For detailed architecture documentation:**
+
+@.claude/ref/architecture/pipeline.md
+@.claude/ref/architecture/modules.md
+@.claude/ref/architecture/cli.md
+
+---
+
+## 🚨 Code Standards
+
+### Quick Reference
+
+- **Comments**: Only when code can't explain itself
+- **Imports**: Always at top of file/module
+- **Test organization**: Use `mod` blocks, not comment dividers
+- **Line length**: Max 100 chars (rustfmt), use `format!()` for long strings
+
+**For detailed standards:**
+
+@.claude/ref/standards/code-style.md
+@.claude/ref/standards/error-handling.md
+@.claude/ref/standards/logging.md
+@.claude/ref/standards/constants.md
+
+---
+
+## 🧪 Testing Standards
+
+### Quick Reference
+
+- **Environment variables**: `ZERV_TEST_NATIVE_GIT`, `ZERV_TEST_DOCKER`
+- **Git operations**: Always use `get_git_impl()`
+- **Docker tests**: Always use `should_run_docker_tests()` gating
+- **Integration tests**: Prefer `TestCommand::run_with_stdin()` (90% of cases)
+
+**For detailed testing patterns:**
+
+@.claude/ref/testing/overview.md
+@.claude/ref/testing/unit-tests.md
+@.claude/ref/testing/integration-tests.md
+
+---
+
+## 🚀 CI/CD
+
+@.claude/ref/workflows/cicd.md
+
+---
+
+## 🛠️ Using Claude Code Features
+
+### When to Use What
+
+**Slash Commands** (Simple, predefined workflows):
+
+- `/audit` - Run code quality audit and fix violations
+- Use for: Detecting and fixing code quality violations efficiently
+
+**Agents** (Complex, multi-step exploration):
+
+- Codebase exploration: "How does version bumping work across the codebase?"
+- Large refactoring: "Migrate all tests to use environment-aware pattern"
+- Research tasks: "Find all API endpoints and document them"
+- Use for: Open-ended tasks requiring autonomous decision-making
+
+**Direct Questions** (Fastest for simple tasks):
+
+- "Review this function for bugs"
+- "Explain how this works"
+- "Fix the error in this code"
+- Use for: Quick reviews, explanations, single-file changes
+
+**Rule of thumb**:
+
+1. Try direct question first (fastest)
+2. Use slash command for standardized workflows
+3. Use agent only for complex multi-file exploration/refactoring