Creative Archive Taxonomy Agent

Studios collect thousands of assets with inconsistent names and tags, so people keep using old or wrong versions. This agent reads the archive metadata, lists ambiguous or untagged files, starting with the most used, and proposes tags from the studio's taxonomy based on file content and context. It writes tags to a staging copy, then runs test searches for common requests and checks whether the right, current asset ranks first. When a search brings up the wrong file, it refines the tags or marks old versions as deprecated and tests again. It never merges files when the evidence about their identity contradicts itself. The librarian approves taxonomy changes and merges before anything goes live. Edge case: two files with the same name but different colors are kept separate and flagged.