Artifacts

Artifacts are the generated outputs from code generation and refinement. Each artifact is a concrete file — a TypeScript module, a Python file, an API route implementation.

Task Types

Inference Tasks

Created by remote_start_inference. Contains the Aexol spec, target language, model, status, and generated artifacts.

From Tasks

Created by AI Import (natural language → Aexol). Contains the natural language input and generated spec.

Refinement Tasks

Created by remote_refine_*. Tracks the parent task, refinement instruction, and refined result.

Task States

Viewing & Managing

Via Studio

• Task list at /studio/tasks — all your generation tasks
• Task detail — view artifact content, metadata, generation parameters

Via MCP

# List tasks
remote_list_inference_tasks  { projectId, limit }

# Get task with artifacts
remote_get_inference_task    { taskId }

# Cancel running task
remote_cancel_inference_task { taskId }

# Wait for completion
remote_wait_task             { taskId, taskType: "inference" }

Batch Artifacts

When generating with a commands array, each command produces its own artifacts within the same task. The Studio uses a fixed, read-only set of four artifacts, in this order:

Task: "Generate e-commerce backend"
├── GRAPHQL  → generated/schema.graphql
├── FRONTEND → frontend.json (routes, components, layouts, RBAC)
├── E2E      → tests/e2e/
└── BACKEND  → src/ (server structure, routes, cron jobs, webhooks)

Next Steps

• Code Generation — Creating artifacts
• Refinement — Improving artifacts
• Fine-Tuning — Better models for better artifacts