Studio

DocsStudioKanban & Project Planning

Plan your project with AI-generated Kanban boards. Generate task backlogs from Aexol specs, manage tickets across BACKLOG→TODO→IN_PROGRESS→DONE, and track execution progress.

Kanban & Project Planning

Turn your Aexol specification into an executable project plan. The Kanban system generates a structured task backlog from your spec — each task targeting a specific entity, workflow, visitor capability, or infrastructure concern — then lets your coding agent execute those tasks one by one.


Overview

The Kanban board lives at /studio/kanban?projectId=your-project. It has four columns:

ColumnMeaning
BACKLOGTasks waiting to be prioritized
TODOReady to work — ordered by priority
IN PROGRESSAgent is currently working on this task
DONECompleted — completedAt timestamp recorded

Each task carries:

  • Title — what needs to be done
  • Description — acceptance criteria, context, technical notes
  • Priority (0–100) — higher = more urgent
  • Tags — comma-separated: auth, api, frontend, testing, devops
  • Assignee — team member responsible

Generating the Project Plan

  1. Open your project in Studio
  2. Navigate to the Generate panel
  3. Select "Project Plan" from the artifact type dropdown
  4. Choose a model (Claude 3.5 Sonnet recommended)
  5. Click Generate

The AI reads your .aexol specification and produces a structured task list.

What the AI analyzes from your spec

Spec elementTasks generated
type definitionsOne task per entity (CRUD, validation, schema)
visitor capabilitiesOne task per permission boundary
workflow stepsOne task per workflow stage
enum + union typesValidation and serialization tasks
Implicit infrastructureAuth, error handling, logging, tests

Priority bands

The AI assigns priorities automatically:

RangeCategoryExamples
90–100Security / AuthLogin, JWT middleware, RBAC
75–89Core EntitiesUser CRUD, Order model, Payment schema
55–74Supporting FeaturesEmail notifications, file upload
35–54UX / PolishForm validation, loading states, error pages
10–34Nice-to-havesDark mode, analytics, admin dashboard

Generation options

When calling via the backend API or Remote MCP, you can configure:

OptionTypeDefaultDescription
granularityepic / story / tasktaskLevel of detail per card
clearExistingbooleanfalseDelete existing backlog before generating
maxTasksnumber30Maximum tasks to generate
defaultColumnBACKLOG / TODOBACKLOGWhere new tasks land
Remote MCP
curl -X POST "https://api.aexol.ai/mcp" \
  -H "Authorization: Bearer sk-aexol-team-..." \
  -d '{
    "jsonrpc": "2.0", "id": 1, "method": "tools/call",
    "params": {
      "name": "remote_start_inference",
      "arguments": {
  "spec": "...",
  "commands": [
    { 
      "artifactType": "PROJECT_PLAN",
      "options": {
        "granularity": "task",
        "maxTasks": 25,
        "defaultColumn": "BACKLOG"
      }
    }
  ]
}
    }
  }'

Managing the Board

In Studio

Drag cards between columns. Click a card to see the full description, edit metadata, or delete.

Agent-Driven Execution

Once your board is populated, the Spectral coding agent can execute tasks autonomously. See the Agent Kanban documentation for the full workflow.

Quick preview of the execution loop:

kanban_next  →  picks highest-priority TODO → IN PROGRESS
    │
    ▼
Agent reads task description, implements (read/write/edit/bash)
    │
    ▼
kanban_move  →  DONE
    │
    ▼
kanban_next  →  next task

Re-planning

When your .aexol specification changes, re-generate the Project Plan. The AI performs a diff-based update:

  • New spec elements → new tasks added to BACKLOG
  • Removed spec elements → existing tasks marked with [OBSOLETE] tag
  • Changed spec elements → task descriptions updated, priority recalculated
  • Existing tasks not in the spec → left untouched (manual tasks)

Enable clearExisting: true if you prefer a fresh board on every generation.