Give an AI coding agent like Claude Code structural context about a codebase before it edits a function.
Ask which other parts of a project call a specific function or what tests cover it.
Generate human-readable wiki documentation automatically from the codebase graph.
Query relationships across multiple registered repositories from one MCP server.
| ontograph/ontoindex | 0xbebis/hyperpay | alfredxw/nova | |
|---|---|---|---|
| Stars | 14 | 14 | 14 |
| Language | TypeScript | TypeScript | TypeScript |
| Setup difficulty | moderate | hard | moderate |
| Complexity | 3/5 | 5/5 | 3/5 |
| Audience | developer | developer | writer |
Figures from each repo's GitHub metadata at analysis time.
Requires Node.js 20 or 22 LTS, graph data stays local, no data leaves the machine.
OntoIndex is a tool that builds a detailed map of a software project and makes that map available to AI coding agents. The problem it addresses is that AI tools typically see only a small piece of a codebase at a time. When an AI edits a function, it may not know which other parts of the project call that function, what tests exist for it, or what breaks if a name changes. OntoIndex pre-computes all of those relationships ahead of time so the AI can ask structural questions before making changes. The map it builds is a graph where nodes represent files, folders, functions, classes, routes, documentation sections, and process steps. Edges between nodes represent relationships: one function calls another, a class extends another, a file imports another, and so on. That graph is stored locally in a folder inside the project directory, so no data leaves the machine. Agents and developers can access the graph through several interfaces. A command-line tool handles indexing, status checks, and wiki generation. An MCP server (a standard protocol for AI tools to call external services) lets AI coding environments like Claude Code or similar tools query the graph directly. An HTTP API and a browser-based interface are also available for more visual exploration or integration with other tools. The workflow is straightforward: run ontoindex analyze to build the graph for a project, then run ontoindex setup to configure it for an MCP-aware AI tool, then start the MCP server. The AI can then ask questions like where a symbol is used, what process a piece of code participates in, or what the impact of a proposed change would be. There is also a wiki generation command that writes human-readable documentation from the graph. OntoIndex supports multiple repositories in a named registry, so an agent working across more than one project can query them in context. Installation is via npm and requires Node.js 20 or 22 LTS. The project is licensed under AGPL-3.0 and includes a disclaimer that it has no associated cryptocurrency.
A tool that builds a local graph of a codebase's files, functions, and relationships, then exposes it to AI coding agents through an MCP server.
Mainly TypeScript. The stack also includes TypeScript, Node.js, MCP.
You can use and modify the code, but if you run a modified version as a network service, you must publish your changes.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.