Skip to content
Foundry

Tasks

What it is

Section titled “What it is”

Tasks are the actionable work items in Foundry. They live on a Kanban board with five status columns and support sprint assignment, subtask decomposition, design context cards, implementation panels, and per-task audit trails.

Why it matters

Section titled “Why it matters”

Tasks connect high-level requirements to concrete deliverables. The board provides instant visibility into progress, while subtask decomposition lets AI agents break complex work into smaller pieces that can be executed in sandboxes.

Key concepts

Section titled “Key concepts”
  • Status columns: Backlog, To Do, In Progress, Review, Done — drag cards between columns to update status
  • Priority: critical, high, medium, low — displayed on each task card
  • Sprint assignment: Tasks can be assigned to a sprint via sprintId; the sprint name appears on the card
  • Workstream scope: Tasks inherit their workstream short code for filtering
  • Subtask panel: Expand a task to see AI-decomposed subtasks with their own status tracking via SubtaskPanel
  • Subtask detail tabs: Drill into individual subtasks with SubtaskDetailTabs
  • Task filters: Filter by status, priority, workstream, sprint, and assignee via TaskFilters
  • Design context: TaskDesignContextCard shows relevant design assets linked to a task
  • Implementation panel: TaskImplementationPanel (under source-control/) connects tasks to code changes
  • Task audit trail: TaskAuditTrail (under audit/) tracks every state change
  • Task verification: Verification workflows confirm task completion quality

How to use it

Section titled “How to use it”
  1. Navigate to [Program] > Tasks.
  2. View the Kanban board with cards organized by status.
  3. Drag a card to a different column to change its status.
  4. Click a card to open the task detail page.
  5. Use Filters to narrow the board by workstream, sprint, or priority.
  6. On the detail page, expand the Subtask Panel to see decomposed subtasks.
  7. Launch a sandbox from a task to execute AI skill work against it.

Data model

Section titled “Data model”

This feature uses the following tables:

  • tasks — Title, description, priority, status, sprintId, workstreamId, assignee
  • subtasks — AI-decomposed sub-items linked to a parent task
  • taskDecompositions — Records of AI decomposition runs
  • taskDesignSnapshots — Design context snapshots injected into task execution
  • taskVerifications — Verification records confirming task quality
Section titled “Related”
Sprints
Sandbox
Agents