Skip to main content

Dionysys documentation

Adaptive UI docs for deterministic and MCP-driven experiments

Learn how the packages score behavior, lock variants, expose React state, and let MCP resources choose UI actions for the Excalidraw demo.

Open DocsPackage Usage
Dionysys
01Interaction eventsshape.drawn + text.added
02Persona evidencedraw_first 0.62
03Decision pathPolicy or MCP action
04UI statetoolbar.allowlist
personaScores{ guided_novice: 0.18, draw_first: 0.62 }

Two decision modes

Pick the adaptation path that fits the experiment.

Deterministic

Inference and policy

Convert telemetry into persona probabilities, then let the contextual bandit policy select the UI variant.

MCP

Resources and actions

Summarize interactions, calculate resource-driven persona scores, and resolve one exposed UI action through a provider-neutral connector.

Documentation map

Everything needed to use and tune Dionysys.

Package Usage

Wire core inference, policy selection, MCP decisions, and React adaptive state.

Configuration

Tune deterministic inference, contextual bandits, and resource-driven MCP mode.

Excalidraw Demo

Edit variants, MCP personalities, toolbar controls, menu items, and connector settings.