Knowledge Layer
When you connect your repositories to ProdE, they go through an indexing process that builds the Knowledge Layer — an AI-enriched representation of your code, including functions, classes, dependencies, cross-repo relationships, and auto-generated documentation.
The Knowledge Layer is what makes everything else in ProdE work. When you ask a question in Codebase Chat, or when your coding agent uses the Codebase Understanding MCP server, the Knowledge Layer is what finds the right code across your repositories and provides accurate, grounded answers.
What Gets Indexed
- Code structure — functions, classes, modules, and their relationships
- Cross-repo dependencies — how services interact across repositories
- Auto-generated documentation — AI-written summaries of your code symbols and modules, including Repo Wikis covering architecture, components, data flows, and more
Connected repositories are synced daily to keep the Knowledge Layer up to date.
How You Use It
You don't interact with the Knowledge Layer directly. Instead, it powers two interfaces:
- Codebase Chat — ask questions about your code in ProdE's web interface
- Codebase Understanding MCP server — bring the same intelligence into your coding agent (Cursor, VS Code, Windsurf, Claude Code)
Getting Started
To build your Knowledge Layer, connect your Git provider and select the repositories you want ProdE to index.