CodeNib

•
112
•

Provides a multi-view data system for delivering comprehensive repository context and typed graph navigation to coding agents.

CodeNib is a sophisticated multi-view data system engineered to provide AI coding agents, such as Claude Code and Codex, with rich repository context and advanced typed graph navigation. It empowers agents to efficiently locate relevant code blocks, trace function calls, and understand complex codebases across a wide array of programming languages. Emphasizing inspectability and reproducibility, CodeNib offers local search capabilities, symbol graph generation, and the ability to export findings as shareable Wikis. This tool significantly enhances an agent's ability to reason about code, demonstrated by its strong performance in code-block recall for real-world issues.

Características Principales

01Inspect code context and navigate callers, callees, definitions, and references across multiple languages
02Supports 14 languages for code chunking and retrieval, with 12 offering full graph backend capabilities
03Export repository context and findings as shareable Wikis with source citations, including GitHub Pages integration
04Utilizes incremental indexing to reuse unchanged views and rebuild only modified code views efficiently
05Integrates with installed agent clients to provide code context without requiring its own model, API key, or GPU
0684 GitHub stars

Casos de Uso

01Generating and sharing browsable documentation (Wikis) directly from codebases for collaboration or reference
02Tracing call stacks and understanding code dependencies across a repository to debug or refactor complex systems
03Equipping AI coding agents with deep source context and typed graph navigation for efficient code understanding and generation