Select a topic under art-topic-model.canon to see cluster controls.
Recognition Types
Matched Nodes
Recognition Models
Generate Models
Quality Level
Min (0 imp.):phrases
Max (100 imp.):phrases
Expand
Direct children
+2 levels
+3 levels
+4 levels
Collapse
Direct children
+2 levels
+3 levels
+4 levels
Select
+1 level
+2 levels
+3 levels
+4 levels
All levels
Files only
Deselect
+1 level
+2 levels
+3 levels
+4 levels
All levels
Clear selection
Select children at level
Level 4
Level 5
Level 6
Check quality
Regenerate topic file
Generate children (+1 level)
Generate children (+2 levels)
Generate topic file (AI)
Extend in active layer
Add child in active layer
Move to active layer
Expand all children in this level
Collapse all children in this level
Generate all children (+1 level)
Generate all children (+2 levels)
Expand all in same level
Collapse all in same level
Generate all in level (+1 level)
Generate all in level (+2 levels)
Fix Strategy Preview
Quality Check & Fix
Processing...
Batch Processing
◼Completed:0
◼Pending:0
◼Failed:0
◼Active:0
API:0
◻Log
0
Server Batch Queues
No active server batches
Create New Layer
#e040fb
Edit Layer
Layer settings and personas
General
Factors 0
Personas 0
Matrices 0
Layer-wide factors for evaluating topics in this domain
Rational Factors
Emotional Factors
No factors defined. Click "Generate" to create layer-specific factors based on your generation prompt, or add them manually.
Layer-specific personas and their factors
Matrix files for this layer's personas. These are created through the normal matrix generation flow when this layer is active.
Loading...
Edit Personas
Base personas.json — factors used for matrix generation
0 personas
Extend Topic
Layer: ...
Add Child in Layer
Create a new topic node as a child
Test Objective Opinion
—
⚙️ Settings
Route every WTM API call through api.mutual.ai's unified auth + cost ledger.
Leave the base empty to fall back to the legacy local server (same-origin).
Local Claude API is recommended for best quality (free with subscription).
Session Costs
No API calls yet
Use OpenAI Batch API for 50% cheaper batch processing. Results ready within 24h.
Topics processed in parallel (local CPU work: pre-fixes, quality checks, partitions).
Max simultaneous AI API calls. Ramps up gently on start to avoid rate limits.
0 = auto-adjust based on provider feedback. Set a value to enforce a hard limit.
Automatically commit and push generated topic files to the Git repository. Disable this if you want to manually control commits.
Automatically generate child nodes when creating a topic file. Children are saved to the world model schema immediately.
Uses a programming-capable AI to classify quality gaps as programmatic, API lookup, or AI generation.
When disabled, falls back to rule-based classification.
Cheap model used for small individual gap fixes within partitions. The main generation model is used for partition regeneration.
Model used for companion framing text and simple tasks. Pick a cheap or free model for low-cost operation.
Template Analytics
Loading topic files...
0 / 00%
0 files failed
No importance matrix loaded. Templates are grouped by topic and sorted alphabetically.