WorkScribe — where project management meets team knowledge

Most teams end up stitching together two separate tools: one for tracking work, another for writing it down. Jira lives in one tab, Confluence in another, and the two rarely talk to each other. WorkScribe started as an experiment in closing that gap — a single workspace where a task and the page documenting it are one click apart.

What it is

WorkScribe is a full-stack, multi-tenant SaaS platform that combines sprint-based project management with a hierarchical team wiki. It's not a toy CRUD app — it's built the way a real product would be: role-based access control, JWT auth with Google OAuth, real-time WebSocket notifications, and strict org-scoped data isolation across tenants.

Core features

Project management

Kanban boards with drag-and-drop across columns
Sprint planning — create, start, and complete sprints
Backlog management with drag-and-drop into and out of sprints
Inline task editing — status, priority, assignee, labels, due dates
Subtasks, rich-text comments, and a full activity log per task
Task-to-wiki-page linking

Team wiki

Hierarchical page tree with drag-and-drop reordering
Rich text editor with headings, tables, code blocks, and slash commands
Autosave with a debounce, plus manual save and an unsaved-changes indicator

Real-time collaboration

WebSocket-powered notifications for task assignment, completion, and @mentions
Notification bell with unread badges

Auth and access control

Email/password and Google OAuth 2.0, with silent refresh token rotation
Owner / Admin / Member roles, enforced per organization
Org invitations delivered by email

Search

A command palette (Cmd+K) for jumping straight to tasks or pages
Full-text search across tasks and wiki content within an org

Under the hood

LayerTechnologyBackendPython 3.12, FastAPIDatabasePostgreSQL 17 (Neon), async SQLAlchemy 2.0CacheRedis (Upstash)FrontendReact 19, TypeScript, ViteStateTanStack Query, ZustandEditorTiptapDeploymentRender (backend), Vercel (frontend)

A few decisions I'm particularly happy with:

Service-layer architecture — routers stay thin; all business logic lives in services/, which made the codebase far easier to test and reason about.
FastAPI BackgroundTasks instead of Celery — no separate worker process to deploy, and it's plenty for email and notification dispatch at this scale.
Redis that fails silently — a cache outage degrades performance, never breaks a request.
Cross-tenant isolation enforced at the service layer — every query is scoped by org_id, so one organization can never see another's data.

Try it

WorkScribe is live at work-scribe.vercel.app, seeded with three real accounts across three organizations — an owner, an admin, and a member — so you can see how permissions actually differ by role rather than reading about it.