Projects

Projects is FanDesk's team project management system — structured work coordination with Kanban boards, sprint planning, GitHub integration, custom fields, and AI that can read, write, and manage tasks on your behalf. Projects are also the foundation for AI-driven workflows: any AI agent can take a project full of well-scoped tasks and execute them autonomously.

Creating a Project

1. Go to Projects in the sidebar (shortcut: 4)
2. Click + New Project
3. Choose a template (or start blank)
4. Name your project and set a key — a short prefix for all task IDs (e.g., "BACK" for Backend, "SALE" for Sales)
5. Add team members and set the project lead
6. Click Create

Project Templates

FanDesk includes templates for different workflows:

Each template comes with appropriate status colors, default custom fields, and workflow settings that match the use case.

Project Views

Kanban Board

Visual columns for each status. Drag tasks between columns to update their status. See task count per column and swimlanes by assignee.

List View

All tasks in a flat scrollable list sorted by priority, due date, or creation date. Best for bulk editing and quick scanning.

Table View

Spreadsheet-like view with all fields visible as columns — assignee, priority, due date, sprint, custom fields. Sort and filter by any column.

Calendar View

Tasks plotted on a calendar by due date. Drag tasks to reschedule. See the full month or week to identify pile-ups.

Gantt View

Timeline view showing task duration, start dates, end dates, and dependencies. Identify the critical path at a glance.

Task Keys

Every task gets a unique, permanent key that follows it everywhere:

PROJECT_KEY-NUMBER

Examples:

• BACK-42 — Backend project, task #42
• MOBILE-1 — Mobile project, first task
• OPS-156 — Operations project, task #156

Task keys are:

• Searchable globally (type a key in search to jump directly)
• Linkable anywhere — paste a key in a message or document and it becomes a clickable link
• Referenced in GitHub commits and PRs (if GitHub integration is enabled)

Custom Fields

Add structured data to tasks beyond the defaults:

• Text fields — Free-form labels, notes, or references
• Number fields — Story points, estimates, scores
• Date fields — Additional dates (start date, review date)
• Select fields — Single-value dropdown (e.g., "Priority", "Team", "Epic")
• Multi-select fields — Multiple values (e.g., "Tags", "Affected Areas")

Custom fields appear in all views and can be filtered and sorted in the Table and List views.

Project Members

Roles

• Lead — Full project control: settings, members, statuses, archiving. Receives auto-assigned unowned tasks.
• Member — Can create, edit, and move tasks within the project.

Adding Members

1. Open project settings
2. Go to Members
3. Search for team members by name
4. Set their role (Lead or Member)
5. Click Add

Project Lead & Auto-Assignment

When a task in your project remains unassigned for more than 24 hours, FanDesk automatically assigns it to the project lead. This ensures nothing falls through the cracks in busy projects.

Project Settings

Sprints

Enable sprints from project settings to work in focused time-boxed iterations. See Sprints & Velocity for full sprint documentation.

GitHub Integration

Link FanDesk tasks to GitHub commits and pull requests:

• When a commit mentions a task key (e.g., "Fix BACK-42: auth timeout"), FanDesk links the commit to the task
• PRs referencing task keys appear in the task's activity log
• Task status can be auto-updated when a PR is merged (configurable)

---

Using Projects for AI-Driven Automation

Projects are the native interface for AI agents to do structured work. When tasks have full context — scope, dependencies, acceptance criteria, assignee — an AI agent can pick them up one at a time and execute them without needing to ask clarifying questions. The more structured your tasks, the more effective AI automation becomes.

Example 1: Sales Outbound Automation

Goal: Automate personalized outbound sales prospecting at scale.

Setup:

1. Create a project called "Sales Pipeline" with the Sales template
2. Customize statuses: Lead → Contacted → Meeting Scheduled → Proposal Sent → Closed
3. Add custom fields: Company, Title, LinkedIn URL, Notes, Next Action Date

How it works:

• Add prospects as tasks (one task per prospect): "Reach out to VP of Engineering at Acme Corp"
• Include full context in the task description: company background, why they're a fit, personalization angle
• Ask DeskMate: "For every Lead-status task in the Sales Pipeline, draft a personalized outreach email based on the task description and contact info, then send it via email"
• DeskMate reads each task, drafts a personalized email using the task context and the knowledge graph (existing relationship history), and asks for approval before sending
• After sending, DeskMate updates the task status to "Contacted" and schedules a follow-up reminder
• The Kanban board becomes your real-time pipeline view: see exactly how many leads are at each stage

Result: Personalized outreach at scale, with every touchpoint tracked in the project board and the knowledge graph enriched with each interaction.

Example 2: AI Code Automation

Goal: Use an AI coding agent (like Claude Code) to autonomously implement an engineering sprint.

The insight: AI coding agents are extremely effective when given well-scoped, context-rich tasks — and extremely wasteful when they have to figure out scope from vague requirements. FanDesk projects give you the structure to make AI coding radically token-efficient.

Setup:

1. Create a project called "Engineering Sprint" with the Engineering template
2. Break each feature into granular tasks with full context. Example task — BACK-47: "Add rate limiting to /api/auth/login":

Scope: Add per-IP rate limiting to the login endpoint.
Files to modify: apps/api/routers/auth.py, apps/api/core/middleware.py
Requirements:
- Max 5 attempts per IP per 15 minutes
- Return 429 with Retry-After header on limit hit
- Use Redis for the rate limit counter
Acceptance criteria:
- Existing login tests still pass
- New test: 6th attempt returns 429
Dependencies: None (Redis already configured)

How it works:

• Give the AI coding agent access to read FanDesk tasks via MCP tools
• The agent reads one To Do task at a time, gets full context from the task description
• Implements the change, writes tests, then updates the task status to In Review
• Moves to the next To Do task
• You review PRs and move tasks to Done

Why this works:

• Full context in each task = agent needs zero clarifying questions = zero wasted tokens
• Granular tasks = each task is one focused change = predictable, reviewable PRs
• Status tracking = you always know exactly what the agent has done and what's next
• Task keys in commits = full traceability from task to code change

---

Next: Learn about sprint planning in Sprints & Velocity.