Document audit trail and compliance reporting

The audit trail records every action taken on every document and cannot be edited from the application — not by a user, not by an administrator, not through the API. Entries are added, never amended or removed. That property is the whole value: a log an administrator can tidy is a log that proves nothing about administrator behaviour, which is exactly the behaviour an auditor is most interested in.

Each entry carries the acting user, the timestamp, the source IP address, the action, the affected document and the specific version it applied to. Version-level precision is what lets you answer the question auditors actually ask, which is not "was this approved" but "was the version in use at that date the approved one".

Views and downloads are logged as well as edits. This is a deliberate choice and it produces a lot of entries, but access is the event that matters in a confidentiality investigation, and a system that logs only changes cannot tell you who read a record before it leaked.

Permission changes, share creations, signature requests, workflow decisions, deletions and restores are all recorded as distinct event types, so filtering the trail is useful rather than an exercise in reading everything.

An audit sample usually means "everything that happened to these twelve documents between these dates", and it usually needs to leave the system as a file that goes in an evidence pack. The trail exports filtered by document, folder, user, action type or date range.

Export is itself an audited action, which sounds recursive but is the point: knowing who took a copy of the access log is part of controlling access to it.

The dashboard reports storage consumption, activity levels, pending signatures and document throughput. These are the numbers that drive day-to-day decisions — whether an approval queue is backing up, whether one team is generating most of the storage growth, whether signature requests are being chased.

Real-time notifications over WebSocket mean counts update as things happen rather than on a page refresh, and users can watch a document or folder to be notified of changes without being a participant in its workflow.

Three reports cover most recurring requests. Storage usage attributes consumption by folder, entity and document type, which is how you answer a finance question about growth with data rather than a shrug. User activity shows what each account has done over a period, which is the starting point for an access review. Duplicate documents lists content-hash matches, which is usually the cheapest storage saving available and, more importantly, finds the places where a superseded version is circulating as current.

A shared-access overview lists everything currently shared and with whom, including external signing links. Shares accumulate silently and are the most common way a permission model degrades; seeing them in one list is what makes them reviewable.