Maple Structured Knowledge Archive

Problem

Engineers often dump weekly reports, meeting notes, incident reviews, and technical notes into folders, then rely on keyword search and memory to recover context. The text may be present, but it lacks topic hierarchy, and code, decision context, and related assets remain scattered.

How It Works

maple-structured-storage turns unstructured documents into a standardized topic archive. It analyzes the input, proposes a short topic name, and creates a topic folder with layered files: description.md, process.md, meta.md, and code.md, while copying referenced resources. After writing, it rebuilds index.json from all topic descriptions and runs a reflection pass that asks up to three follow-up questions with source snippets to fill missing people, time, cause, related systems, or decision context. Search reads index.json first, then opens the most relevant files based on the requested level of context.

Boundaries

It is best for archiving existing .txt / .md documents, weekly reports, postmortems, and technical notes, not for real-time collaborative knowledge graphs. It prioritizes preserving information, keeping code intact, and appending without dropping prior content, so it behaves more like an archive and review workflow than an automatic summarizer. Confirm source_path and storage_path before first use, and note that the index is rebuilt fully rather than incrementally.