Docs/Documentation

WORKX Docs

Everything you need — from first login to AI-assisted requirements, real-time collaboration, workflow automation, and full audit logging.

Feature Overview

What is WORKX?

WORKX is an AI-native project and team management platform for software product teams. It combines:

  • Structured requirements & tasks with real-time collaborative editing (Tiptap + Yjs)
  • Granular RBAC at org and project level — Owner, Admin, Member, and custom roles
  • BYOK AI — bring your own OpenAI, Anthropic, or Gemini keys; stored encrypted
  • Visual workflow automation with a ReactFlow node editor
  • Full audit logging with GitHub-style heatmaps and per-user activity timelines
  • MCP integration for connecting external AI agents (Claude Desktop, etc.)

Sign in

1

Go to the login page

Visit /login. Sign in with Google OAuth.

2

Authenticate with Google

Click Continue with Google. After approval you are redirected to your org, or prompted to create one.

3

Session tokens

WORKX issues a short-lived JWT access token and a rotatable refresh token, stored as HttpOnly cookies (hq_access / hq_refresh).

Create an organization

An organization is your team workspace. All projects, members, settings, and billing live inside an org.

1

Open the New Org wizard

Click + New org in the sidebar. A multi-step wizard walks you through: name → database type → storage → AI provider → member invite.

2

Choose a database type

Three options: Platform Shared (quickest), BYODB (connect your own Postgres), or Dedicated Managed (plan-gated).

3

Invite founding members

The wizard lets you send email invites immediately, or skip and do it later from Settings → Members.

Create a project

1

Click + New project

Inside your org, go to Projects and click + New project. Give it a name, key (e.g. SHIP), icon, and description.

2

Choose a view

Each project has two task views: Kanban (drag-and-drop columns) and List (flat sortable list).

3

Explore project tabs

Projects have tabs: Overview · Tasks · Requirements · Knowledge · Members · Settings. Each tab is permission-gated via RBAC.

Invite team members

1

Go to Settings → Members & Invites

In your org sidebar, click Settings, then the Members & Invites tab.

2

Invite by email

Click Invite User. Enter their email and assign an org-level role. They'll receive an email invite link.

3

Project-level access

Inside a project, go to the Members tab to add members and assign a project-specific role that overrides their org role.

Task Management

Tasks are the atomic unit of work. Each task can have a rich-text document body, comments, assignees, due date, and priority.

1

Create a task

In the List or Kanban tab, click + New Task. Set a title, status (TODO / IN_PROGRESS / DONE), and assignee.

2

Add a document body

Open a task and click the document tab. A full Tiptap editor lets you add rich content — the same editor used for requirements.

3

Comments & filters

Thread comments directly on a task. Filter the task list by status, assignee, project, or date from the filter bar.

⚠️
Tasks are write-locked if your org's database is undergoing a migration. A banner at the top of the page will indicate this state.

Requirements Management

Requirements are structured specifications with a full lifecycle: DraftReviewApprovedReleased.

1

Create a requirement

In the Requirements tab, click + Create Requirement. It opens as a full collaborative document.

2

Add structure

Add: Acceptance Criteria, Comments, Attachments (files/images via org storage), a Decision Log, and linked Research docs.

3

Version and release

Click Release to create an immutable snapshot. Compare any two versions side-by-side in the full-screen diff modal.

Collaborative Editor

Requirement and task editors use Tiptap + Yjs (CRDTs) for real-time multi-user collaboration.

1

Multiple editors, one document

When multiple users open the same document, their cursors and text selections appear in real-time with unique colours shown in the presence bar.

2

Inline AI

Type /// anywhere in the editor to open the AI inline toolbar. Presets: Improve prose, Summarise, Generate acceptance criteria, Translate.

3

Mermaid diagrams

Insert a Mermaid diagram block from the editor toolbar. Diagrams are syntax-validated and rendered live inside the document.

🔌
The collab backend runs as a separate BACKEND/ws Yjs server. Short-lived collab tokens are issued per-session and verified on WebSocket upgrade.

Roles & Permissions (RBAC)

WORKX uses a two-tier RBAC system — org-level roles with optional project-level overrides.

Owner

Full control over the org, billing, and all resources. Cannot be removed.

Admin

Can manage members, roles, projects, and all settings.

Member

Can view and edit projects they're added to. Default for new invites.

Custom roles

Create roles with fine-grained permission sets in Settings → Roles & Permissions.

Project-level roles override org-level roles. Go to Settings → Roles & Permissions to see all 50+ permission keys.

AI Key Setup (BYOK)

WORKX supports Bring Your Own Key — your LLM keys are stored encrypted (AES-256-GCM) and never returned in plaintext.

1

Go to Settings → AI Provider

Find the AI Provider Keys (BYOK) section. Click + Add Provider Key.

2

Choose provider and paste key

Select your provider (OpenAI, Anthropic, Google Gemini), paste your API key, and save.

3

Credential hierarchy

Keys resolve in priority order — the most specific key wins:

Personal key (user override)
    ↓ Role-level key (via role AI policy)
        ↓ Org-level key (shared default)
            ↓ Platform stub (fallback)
✏️
Use the /// command inside any requirement editor to trigger inline AI — improve prose, generate acceptance criteria, or summarise — without leaving the document.

Visual Workflow Editor

Workflows are org-scoped automation pipelines built with a ReactFlow visual canvas.

1

Create a workflow

Go to Workflows in the org nav. Click + New Workflow.

2

Build with nodes

Drag nodes from the palette and connect them. Node types: Trigger, AI, Condition, HTTP, Delay, Code.

3

Publish & run

Save as a draft, then Publish a version and set it as Active. Execution history is stored per workflow.

Knowledge Base

Add reference material that the AI uses as context when generating or improving content.

1

Org-level knowledge

Go to Settings → Knowledge to add org-wide items (text, file upload, or URL fetch).

2

Project-level knowledge

Inside a project, go to the Knowledge tab. Upload files, paste a URL, or add plain text.

3

Use in AI chat

In the AI Chat panel, click Attach Knowledge to inject specific items into the current AI conversation context.

MCP Integration

The Model Context Protocol (MCP) lets external AI agents connect to your org's data through a Bearer-token API.

1

Create an MCP Server Token

In Settings → MCP Tokens, click Create Token. Give it a name (e.g., “Claude Desktop”).

2

Configure your AI client

Copy the generated token and configure it as the Bearer token in your MCP client. The agent can then read projects, requirements, and tasks within your RBAC scope.

3

Security & revocation

Tokens respect RBAC — an agent using a Member token cannot access Owner-only resources. Revoke any token instantly from Settings.

📖
Complete MCP Setup Guide
Step-by-step instructions for Claude Desktop, available tools reference, security tips, and troubleshooting.
View MCP Server Guide

Insights & Audit Log

WORKX logs every action automatically. View them in the Insights page (requires org.audit.view permission).

1

Activity heatmap

A GitHub-style full-year heatmap shows activity density per day. Select a year on the right. Hover a cell to see a breakdown by category. Timezone is adjustable.

2

Audit log feed

A paginated, cursor-based list of all actions. Filter by user, project, category, date range, and role.

3

Per-user activity

Click a team member card (or go to People) to see a full per-user heatmap + timeline with event breakdowns.

Storage Setup

File uploads (requirement attachments, knowledge files) use your org's configured storage provider.

  • AWS S3 — bucket + region + access key
  • Cloudflare R2 — S3-compatible, zero egress
  • MinIO — self-hosted S3-compatible
  • Local disk — for local / container deployments

Configure in Settings → Storage. Click Test Connection before saving to validate bucket access.

Database & Data Plane

Three database configurations per org:

Platform Shared

Quickest to start — data lives in WORKX's Postgres. No config needed.

BYODB

Bring Your Own Database — connect your own Postgres. Connection string is validated before saving.

Dedicated Managed

A dedicated Postgres provisioned for you. Plan-gated. Live migration progress shown.

Configure in Settings → Database. During a migration, write operations are temporarily locked across the org.

Org Settings

Access all org configuration at Settings in the sidebar. Nine tabs:

General

Org name, description, danger zone (delete org).

Members

Member list, invite, manage roles, remove members.

Roles & Permissions

Custom role CRUD, assign permission sets, per-role AI policy.

Database

Storage type (shared/BYODB/dedicated), migration management.

Storage

S3 / R2 / MinIO / local storage config and connection test.

AI Provider

Org-level BYOK AI keys — set default, rotate, revoke.

AI Templates

Org-level AI prompt templates for consistent AI outputs.

My AI Keys

Personal user-level AI key overrides (highest priority).

MCP Tokens

Issue and revoke long-lived tokens for external AI agents.