Endure

Endure is an AI code maintenance intelligence tool for working with git repositories. It is built to help identify files that are likely to cause problems before they do, using information from a codebase’s own history rather than generic rules.

The product scores files by complexity, churn, staleness, and historical risk, then combines those signals into a Debt Score and an overall Health Score for the repository. Its dashboard includes an overview of codebase health, a file view that ranks files by actual risk and allows sorting by debt score, complexity, or churn, a co-change graph that shows files that frequently change together, a knowledge base of recurring failure patterns learned from the repository’s git history, and a trends view that stores each analysis as a snapshot so changes over time can be reviewed. For high-debt files, Endure uses an LLM to infer what a file is supposed to do, stores that intent, and attaches a confidence score, severity, and a CLI command for manual verification.

It accepts either a public GitHub URL or a local repository path and returns a health score, severity distribution, and the highest debt files in under 60 seconds. The co-change graph is derived from git history, and the analysis is intended to run alongside an existing workflow without requiring manual annotations or a new process. The page also describes it as a real, working product in research preview and says it is actively seeking feedback.

The interface shown on the page includes Overview, Files, Co-change Graph, Knowledge Base, and Trends views. No pricing or licensing terms are stated in the provided material, and no maker company is identified beyond the site branding and URL.