ima Knowledge Base Sync Tool

What It Solves

Local folders and ima knowledge bases often drift apart: new or modified files stay local, while some entries exist only in the knowledge base. Manually checking uploads is slow, and it is easy to attempt files that ima does not support or that exceed size limits. This skill treats the local directory, the ima knowledge base, and the sync state as one comparison workflow, which is useful for ongoing knowledge-base maintenance rather than one-off file transfers.

How It Works

The tool is organized around diff states:
- It lists knowledge-base files via ima OpenAPI or Bearer Token
- It scans the local directory recursively and matches local filenames, including extensions, against ima titles
- It uses sync_state.json to record file size and modification time for incremental sync
- It pre-checks supported formats and limits, such as .pdf, .docx, .md, and .png, and automatically flags unsupported video types
- It generates an HTML report with missing, new, modified, and same
- Uploads follow create_media, direct COS upload, and add_knowledge

It can run as a browser UI, a Python script, or a WorkBuddy scheduled check. For automation, a --dry-run pass is a safer starting point.

Boundaries And Caveats

Use this for local folder to ima knowledge base syncing and reporting; search, note creation, or editing inside ima belong to other ima capabilities. File names must match ima titles. Keep OpenAPI credentials in environment variables or trusted UI fields, and treat sync_state.json as important: deleting it can cause repeated uploads.