Skip to content
AssureSwarm Docs

Agent Operating Guide

/coach-schema-design

Interview you about your domain and produce a complete schema JSON in bulk-import format for an admin to paste into the Bulk Import page: never auto-applied.

Drive the onboarding interview that turns “what does your team track” into a complete schema: the item types you need, their fields, shared option lists, and the relationships between them. The skill walks you through your domain one question at a time, then writes a complete JSON in the platform’s bulk-import format. Schema is an admin-page concern, not a data change: nothing routes through suggest_change; you apply the result by pasting it at the admin Bulk Import page (/en/admin/bulk-import). It validates the file as a hard gate before printing those steps, so what you paste is known-good.

When to use

Section titled “When to use”

When you’re setting up a new tenant and need to work out what item types and fields to configure: the hardest part of onboarding. If your work is a standard domain (audit, SOX, regulatory, or GRC), skip the interview and apply a canned /coach-starter-pack instead; the skill offers to short-circuit to one if it recognizes your domain.

Example

Section titled “Example”

/coach-schema-design walks you through your domain one step at a time, what your team tracks, who uses Canvas, which systems you run, proposes item types with sensible icons and colors, builds each type’s fields, then captures shared option lists and the relationships between types. It writes the bulk-import JSON to .coworkcanvas/schemas/<domain>-<date>.json, validates it, and prints the three steps to paste it at /en/admin/bulk-import.

Good to know

Section titled “Good to know”
  • Systems-grounded: before designing types, it asks which systems you actually run: ERP, cloud, identity, security, HR, and so on, and tailors the starter controls, risks, and policies to that stack, emitting only what applies and skipping templates for systems you don’t run. It saves your answers to .coworkcanvas/systems-profile.json and reuses them on re-run.
  • Baseline-aware: if a compliance baseline is already imported (your control and risk types are populated), it doesn’t redesign or duplicate: it adds only the system-specific delta plus the scoping fields.
  • The schema is a starting point, not a final answer. The JSON is meant to be checked into version control and iterated on over weeks as your tenant matures.
  • Re-running is idempotent: it asks whether to refresh the existing file or start fresh.
Section titled “Related”