How To: Use Cognitive Artifacts

Generate and use CADSL-based cognitive artifacts (checklists, decision trees, kanban boards, mind maps, timelines) to support knowledge work in SEA-Forge™.


Prerequisites


Workflow

1. Identify Need

Determine if a cognitive artifact would help:

Ask: “Would visualizing this help?”

Situation Suggested Artifact
Multiple tasks to track Checklist
Choosing between options Decision Tree
Workflow with stages Kanban Board
Exploring concept relationships Mind Map
Planning over time Timeline

2. Generate CADSL Definition

Use the cognitive-artifacts-builder skill:

1
"Create a checklist for our production deployment process"

AI agent will generate CADSL YAML definition with:

3. Review and Refine

Iterate on the artifact:

1
2
3
"Add a critical priority flag to the database backup task"
"Link this to the sea:QualityGate concept"
"Add a progress bar"

4. Integrate with Case

Link artifact to CMMN case:

1
2
3
4
5
case:
  id: "case-2026-001"
  artifacts:
    - type: "Checklist"
      artifactId: "checklist-001"

5. Track Interactions

Artifact usage is tracked via OpenTelemetry:


Examples

Deployment Checklist

1
"Create a checklist for deploying the CADSL renderer to production"

Result: Checklist with pre-deploy, deploy, and post-deploy sections, each with specific tasks linked to relevant concepts.

Architecture Decision Tree

1
"Create a decision tree for choosing the right bounded context structure"

Result: Decision tree guiding through questions about business capabilities, domain logic, and infrastructure concerns.

Sprint Kanban

1
"Create a kanban board for the current sprint with columns for Backlog, In Progress, Review, and Done"

Result: Kanban board with WIP limits, drag-and-drop enabled, task cards with assignees and tags.


Best Practices

  1. Semantic Anchoring: Always include semanticRefs linking to Knowledge Graph concepts
  2. Provenance: Track who created artifact and when
  3. Actionability: Include buttons, links, or commands for next steps
  4. Iteration: Start simple, refine based on usage