Verify a claim ● TrustNode live ● TrustNode unreachable
Run any AI-generated claim through the TrustNode engine — sources, evidence, contradictions, provenance. Fully deterministic: no model grades its own homework.
Earned trust
A source's measured track record. This is what the score is built on.
Community trust
Labeled votes. Displayed, never silently merged into the score.
Personal trust
Your hierarchy — allowlists, blocklists, weights. You decide.
Try:
How this score was computed — show the working
⚠ Conflicts detected
Evidence chain — sources retrieved
No claim verified yet. Paste a claim above — or try a demo — to see the full chain: verdict, conflicts, evidence.
Response Behavior
Shape the style, creativity, and structure of AI responses.
Tone
Quick Modes
Output Format
RAG & Sources
Control what information is retrieved, weighted, and cited.
Source Priority & Weighting
Retrieval Settings
Rerank Results
Reorder retrieved chunks before generation.
Parameters
Control model sampling and output limits.
Stop Sequences
Bias & Safety
Control balancing, filtering, and answer safeguards.
Bias sliders are prompt-level guidance, not a model guarantee.
Preset Manager
Save, manage, and share behavior/source configurations.
Presets are synced locally in this browser — DB-backed sync is Phase 2.
+
Create New Preset
Active Sources
Enabled sources currently shaping responses.
Backend Status ● Live ● Offline (local preview)
Ask calls the backend /ask endpoint at the configured API base URL. When the backend is unreachable, the UI falls back to the local preview.