LinkedIn Wizard — capture, generate, collaborate, publish
LinkedIn Wizard (lk-wiz) is a multi-user web application for turning raw ideas into published LinkedIn posts. Capture an idea from anywhere, generate a draft with GenAI in your voice, refine it collaboratively in a real-time canvas editor, move it through a review pipeline, and schedule it for publication via Zoho Social.
New here? Start with the Getting Started overview for a tour of the pipeline, or jump to the Architecture overview to see how the system fits together.
What lk-wiz does
- Never lose an idea. Capture post ideas from the app, a bookmarklet, Raycast, or a token-based Quick Capture API.
- Generate LinkedIn-quality drafts in seconds using Amazon Bedrock Nova Pro, tuned with your prompts, target audience, language, and best-performing example posts.
- Collaborate in real time. Multiple editors, live cursors, and conflict-free editing through a TipTap + Yjs canvas over a WebSocket API.
- Manage the full pipeline — idea → draft → review → scheduled → published — on a single Kanban board, with versioning and auto-proofreading.
- Publish via Zoho Social. lk-wiz owns the content and scheduling; Zoho Social handles the LinkedIn API.
Who it's for
| Role | What they do |
|---|---|
| Content Creators (Authors) | Capture ideas, generate AI drafts, edit in the canvas, drive their post pipeline. |
| Reviewers & Editors | Review submitted posts, approve or request revisions, co-edit live. |
| Workspace Owners / Team Leads | Manage workspaces, invite members, request assets, oversee the pipeline. |
| Admins | Approve new signups, manage users and roles, monitor system health. |
Documentation map
- Getting Started — the idea-to-publish pipeline, core concepts, and workspaces & roles.
- Architecture — system components, authentication, GenAI, and real-time collaboration.
- API Reference — the HTTP and WebSocket API surface.
- MCP / AI — use lk-wiz from Claude Code and Claude Desktop via the Model Context Protocol.
- Deployment — infrastructure stacks, CI/CD, and how the docs site itself is hosted.
Tech at a glance
| Layer | Technology |
|---|---|
| Frontend | Next.js 14 (static export) + TypeScript + Tailwind + shadcn/ui |
| Backend | Python 3.13 Lambda + API Gateway HTTP + WebSocket |
| Database | DynamoDB (single-table, KMS-encrypted, PITR) |
| GenAI | Amazon Bedrock Nova Pro (amazon.nova-pro-v1:0) |
| Auth | AWS Cognito (Google social + email/password, reCAPTCHA, admin approval) |
| Editor | TipTap + Yjs (CRDT over WebSocket) |
| Notifications | Slack SDK |
| Scheduling | Amazon EventBridge (daily cron) |
| IaC | OpenTofu / Terraform with copebit modules |
| Region | eu-central-1 (separate dev and prod accounts) |