ingest

You can add project documentation to persistent AI memory by ingesting long-form content like pasted text, files, and folders into the Neuralscape system. This allows your AI assistant to automatically reference architecture specs and coding conventions in future tasks.

Yes, you can ingest PDF and MS Office files directly. The system handles server-side parsing for docx, xlsx, and pptx files, along with zips and folders, breaking the content down for accurate recall.

File ingestion works by chunking content into verbatim passages and distilled graph facts. This dual storage method maximizes recall accuracy when the AI agent queries the memory in future sessions.

Yes, persistent memory ingestion works seamlessly in both Claude Code and Claude Cowork environments. It also provides fallback options for scenarios where direct file uploads are unsupported.

The best way to avoid re-sharing context is to ingest your entire project documentation folder into persistent memory. This stores the external knowledge with artifact provenance tracking, making it immediately available across all future sessions.

While the system supports binary documents like PDFs and MS Office files through server-side parsing, unsupported file upload scenarios may require fallback options to successfully ingest the content into persistent memory.