intersense

Project domain classification works by scanning directories, manifests, dependencies, and keywords to assign a primary domain like web-app or ML-pipeline. This mechanism enables targeted review by matching identified domains with specific research directives and criteria.

Yes, you can automatically detect if a project is an ML-pipeline or a web-app by running the provided Python scripts. These scripts analyze project signals from manifests and dependencies to determine the predefined domain automatically.

Project classification does not require external dependencies to run, as the metadata indicates zero dependencies. It utilizes standalone Python scripts for scanning, classification, and content-addressable hashing for staleness detection.

The best way to target research directives based on project type is to classify the project first using automated signals from directories and keywords. Once classified, the system provides domain-specific criteria and research directives tailored to that exact project type.

Content-addressable hashing detects staleness by generating unique hashes based on project file contents during classification. If files change, the hash updates, signaling that the previous domain classification requires re-evaluation to maintain accurate review criteria.