How does this skill help save tokens?

Instead of injecting full source files into the prompt, this skill uses a hierarchical approach that injects only relevant project essences and module states, typically reducing context size from thousands of tokens to approximately 200.

What programming languages are supported?

The tool supports a wide range of languages including C/C++, Python, JavaScript/TypeScript, Go, Rust, Java, Ruby, and C# via Tree-sitter parsers.

What is the difference between Phase 1 and Phase 2?

Phase 1 uses a C++ tool for objective structural parsing (symbols and hierarchy), while Phase 2 uses Claude to add architectural SSL patterns that explain the 'why' and the relationships between components.

Does it require a full re-index whenever I change a file?

No, it uses provenance tracking and hashes to identify 'maybe_stale' files, allowing for incremental updates that only re-analyze modified parts of the codebase.

Codebase Knowledge Mapper

Name: Codebase Knowledge Mapper
Author: genomewalker

bygenomewalker

開発者ツール

Maps and analyzes codebase architecture using tree-sitter AST extraction and semantic relationship patterns for efficient AI context.

概要

This skill provides a powerful two-phase approach to codebase understanding, combining automated C++ tooling with Claude's architectural reasoning. By extracting symbols, hierarchies, and semantic relationships (SSL), it creates a semantic memory substrate that allows for highly efficient, token-saving context injection. It goes beyond simple parsing to capture the 'why' and 'how' of design decisions, making it an essential tool for navigating large-scale projects where traditional full-file context is too expensive or noisy.

主な機能

Incremental updates via provenance and staleness tracking
Automated AST extraction using Tree-sitter for 10+ major languages
Hierarchical state management (Project Essence, Module State, Pattern State)
0 GitHub stars
Relational triplet mapping between files, symbols, and scopes
Soul Semantic Language (SSL) patterns for architectural intent

ユースケース

Optimizing LLM token usage by injecting specific hierarchical context instead of raw files
Rapidly onboarding to a large, unfamiliar codebase by mapping structural relationships
Maintaining a living architectural map that automatically updates as code evolves

概要

主な機能

Incremental updates via provenance and staleness tracking
Automated AST extraction using Tree-sitter for 10+ major languages
Hierarchical state management (Project Essence, Module State, Pattern State)
0 GitHub stars
Relational triplet mapping between files, symbols, and scopes
Soul Semantic Language (SSL) patterns for architectural intent

ユースケース

Optimizing LLM token usage by injecting specific hierarchical context instead of raw files
Rapidly onboarding to a large, unfamiliar codebase by mapping structural relationships
Maintaining a living architectural map that automatically updates as code evolves