Normalize

Structural code intelligence for humans and AI agents.

Status: Potentially Mature ●

v0.3.2. Core functionality is solid: the three primitives (view/edit/analyze) ship, with extensive language support (98 languages via tree-sitter), a background indexing daemon (Unix), shadow-git edit tracking, and AI-session analysis. Active development is on the analysis surface (budget, test-ratio, coupling clusters, output formatting). Remaining work is capability expansion rather than foundation building.

Normalize provides tools for understanding, navigating, and modifying code at a structural level (AST, control flow, dependencies) rather than treating code as text.

Key features

• 98 Languages - Tree-sitter grammars for comprehensive language support
• Structural Editing - AST-based code modifications with fuzzy matching
• Background Indexing - Daemon maintains symbol/call graph index
• Shadow Git - Automatic edit history tracking in .normalize/shadow/
• Session Analysis - Parse and analyze AI agent logs (Claude Code, Gemini CLI, etc.)

Three primitives

Command	Purpose	Example
view	See structure	normalize view src/ normalize view MyClass
edit	Modify code	normalize edit src/foo.rs/func --delete
analyze	Compute metrics	normalize analyze --complexity

Crates

Core

Crate	Description
normalize	CLI binary and main library
normalize-core	Foundational traits (Merge, etc.)
normalize-derive	Procedural macros

Language support

Crate	Description
normalize-languages	98 languages via tree-sitter grammars
normalize-surface-syntax	Syntax translation between languages (TS ↔ Lua ↔ Python)

Code analysis

Crate	Description
normalize-rules	Syntax-based linting with tree-sitter queries
normalize-tools	Unified interface for external tools (oxlint, ruff, prettier, etc.)

Type generation

Crate	Description
normalize-typegen	Polyglot type/validator generation from schemas (incl. JSON Schema)
normalize-openapi	OpenAPI client codegen

Package management

Crate	Description
normalize-packages	Package registry queries for 12+ ecosystems

Agent support

Crate	Description
normalize-chat-sessions	Session log parsing (Claude Code, Gemini CLI, OpenAI Codex)
normalize-session-analysis	Session cost/usage/model metrics
normalize-cli-parser	Parse CLI --help output

Quick start

git clone https://github.com/rhi-zone/normalize
cd normalize
cargo build --release

# View a file's structure
normalize view src/main.rs

# Analyze codebase health
normalize analyze --health

# Search for a symbol
normalize view SkeletonExtractor

Relationship to Moonlet

Normalize focuses on intelligence (code analysis, session parsing, understanding) while Moonlet handles agency/execution (LLM calls, memory, running agents). The projects are intentionally not hard-linked. Normalize can optionally extend Moonlet via a plugin architecture.

Links

• GitHub
• Documentation