Search

PythonReactSEOPDF
GitHub Skill

cognitive-artifacts-builder

Build CADSL-based cognitive artifacts for SEA-Forge™ including checklists, planners, decision trees, kanban boards, and mind maps. Use for generating interactive, semantically-anchored artifacts that support knowledge work and collaborative decision-making. Aligns with Cognitive Architecture patterns and Knowledge Graph integration.

GodSpeedAI☆ 1 Repository stars0 Repository forks

Installation

Follow the repository instructions for this skill and your agent. Review the files and dependencies before adding it to your workflow.

View source ↗Guides

Skill instructions

Cognitive Artifacts Builder

Generate CADSL (Cognitive Artifact DSL) artifacts for SEA-Forge™ to support knowledge work. CADSL artifacts are declarative, semantically-rich, and platform-agnostic definitions that can be rendered across multiple interfaces (web, mobile, voice).

For reference: Cognitive Architecture Handbook | CADSL Specification


When to Use This Skill

Use cognitive-artifacts-builder when:

  1. Organizing Knowledge: User needs structured representation of complex information
  2. Supporting Decisions: User is evaluating options or making strategic choices
  3. Tracking Work: User needs to manage tasks, projects, or workflows
  4. Facilitating Collaboration: Team needs shared artifact for alignment
  5. Visualizing Relationships: User wants to see connections between concepts

Artifact Type Catalog

1. Checklist

Purpose: Track completion of sequential or independent tasks

Elements: Heading, Checkbox, Label, ProgressBar

Use Cases:

  • Pre-flight checklists before deployment
  • Quality assurance verification
  • Onboarding task lists
  • Compliance requirements

CADSL Example: ```yaml artifactType: Checklist metadata: title: "Production Deployment Checklist" semanticRefs: - conceptId: "sea:Deployment" - conceptId: "sea:QualityGate" provenance: generatedBy: "cognitive-artifacts-builder" timestamp: "2026-01-06T12:00:00Z"

sections:

  • id: "pre-deploy" heading: "Pre-Deployment" items:
    • id: "test-pass" checkbox: true completed: false label: "All tests passing"
    • id: "review-complete" checkbox: true completed: false label: "Code review approved" ```

2. Decision Tree / Planner

Purpose: Guide users through conditional decision-making paths

Elements: Heading, Section, RadioButton, Button, Arrow, Conditional

Use Cases:

  • Architecture decision records (ADR)
  • Troubleshooting guides
  • Policy routing (which team handles this?)
  • Feature selection wizards

3. Kanban Board

Purpose: Visualize workflow states and move items through stages

Elements: Panel, Section, DragAndDropArea, Card, Badge, ProgressIndicator

Use Cases:

  • Sprint planning
  • Case management (CMMN artifact pipeline)
  • Content workflow (draft → review → published)
  • Support ticket triage

4. Mind Map / Concept Map

Purpose: Visualize hierarchical or networked relationships between concepts

Elements: Canvas, Node (circle/box), Line, Arrow, Label, Grouping

Use Cases:

  • Brainstorming sessions
  • System architecture overview
  • Knowledge Graph visualization
  • Dependency mapping

5. Timeline / Roadmap

Purpose: Chronological representation of events, milestones, or plans

Elements: Timeline, Calendar, Schedule, ProgressBar, Badge

Use Cases:

  • Project roadmaps
  • Release planning
  • Historical event tracking
  • Compliance deadline tracking

Workflow: Generating CADSL Artifacts

Step 1: Analyze Context

Extract from user conversation:

  • Task/Goal: What is the user trying to accomplish?
  • Information Structure: Linear? Hierarchical? Networked? Temporal?
  • Interaction Needs: View-only? Editable? Collaborative?
  • Semantic Context: Related concepts from Knowledge Graph
Step 2: Recommend Artifact Type

Based on analysis, suggest most appropriate artifact type:

User Need Recommended Artifact
Track completion of tasks Checklist
Choose between options Decision Tree / Planner
Manage workflow stages Kanban Board
Explore concept relationships Mind Map
Plan over time Timeline / Roadmap

Explicitly ask: "Would a [ARTIFACT_TYPE] help with [TASK]? Here's what it would include..."

Step 3: Generate CADSL Definition

Create YAML artifact following specification:

Required metadata: ```yaml metadata: title: "" semanticRefs: - conceptId: "sea:ConceptId" # Link to Knowledge Graph provenance: generatedBy: "cognitive-artifacts-builder" timestamp: "<ISO-8601>" ```

Semantic Anchoring:

  • Every artifact MUST include `semanticRefs` linking to Knowledge Graph concepts
  • Use `conceptId` format: `sea:<ConceptName>` or `:<ConceptName>`
  • Enables discoverability, traceability, and semantic reasoning
Step 4: Present Artifact
  1. Show CADSL definition in code block
  2. Explain structure: Walk through sections and key elements
  3. Highlight semantic links: Point out Knowledge Graph connections
  4. Suggest interactions: "You can edit...", "I can update...", "This links to..."
Step 5: Iterate

Support refinement:

  • "Add a new column to the kanban board"
  • "Change this checklist item priority to critical"
  • "Link this decision tree node to ADR-034"

Use precise updates: Modify specific sections of YAML rather than regenerating entire artifact.


Integration Points

Knowledge Graph (DomainForge™)

Purpose: Semantic anchoring and concept resolution

Pattern: ```yaml semanticRefs:

  • conceptId: "sea:BoundedContext" uri: "http://sea-forge.org/ontology#BoundedContext" ```

Query Knowledge Graph to:

  • Validate `conceptId` exists
  • Retrieve related concepts for suggestions
  • Check SHACL constraints on artifact structure
Case Management (CMMN)

Purpose: Artifacts as first-class case items

Pattern: ```yaml metadata: caseId: "case-2026-001" stage: "planning" # CMMN stage ```

Integration:

  • Artifacts move through CMMN artifact pipeline stages
  • PM-Agent can trigger artifact generation
  • Artifact state changes emit events to NATS
Observability (OpenTelemetry)

Purpose: Track artifact usage and effectiveness

Telemetry:

  • Artifact generation events
  • User interactions (edits, completions)
  • Semantic reference lookups

Quality Guidelines

1. Semantic Richness

❌ Bad: Generic labels without context ```yaml heading: "Items" items:

  • label: "Task 1" ```

✅ Good: Descriptive, semantically linked ```yaml heading: "Pre-Deployment Quality Gates" semanticRefs:

  • conceptId: "sea:QualityGate" items:
  • label: "All tests passing (CI/CD)" semanticRefs:
    • conceptId: "sea:ContinuousIntegration" ```
2. Provenance Tracking

Always include:

  • `generatedBy`: Tool/agent that created artifact
  • `timestamp`: When artifact was created
  • `sourceDoc`: Optional reference to originating specification
3. Actionability

Artifacts should enable action, not just display:

  • Buttons with `command` or `link` properties
  • Interactive elements (checkboxes, drag-and-drop)
  • Clear next steps

References

Examples

Read the usage examples in the source repository.

View source ↗

Source history

Indexed on
2026-01-20
Index updated
2026-01-26
Branch
main

Index and sync dates describe this directory. View the source repository for code history.

GitHub ↗

chat Comments (0)

chat_bubble_outline

No comments yet. Be the first to share your thoughts!