This project was created for the DeepThought Fellowship recruitment assignment.
It demonstrates how psychological reflection can be transformed into a deterministic decision-tree product without using LLMs at runtime.
DT-Daily-Reflection-Tree/
├── tree/
│ └── reflection-tree.json
├── agent/
│ └── app.py
├── transcripts/
│ ├── persona1.md
│ └── persona2.md
├── write-up.md
└── README.md
Part A – Deterministic Reflection Tree
The reflection system guides users through three psychological axes:
Axis 1: Locus
Victim ↔ Victor
Axis 2: Orientation
Entitlement ↔ Contribution
Axis 3: Radius
Self-Centric ↔ Altrocentric
The tree uses fixed options, deterministic branching, reflection nodes, bridge nodes, and a closing summary.
Part B – Runnable Agent
A Streamlit-based reflection tool that:
Loads the tree from JSON
Walks the user through nodes
Branches deterministically
Displays reflections
Produces a final summary
How to Run
cd agent
streamlit run app.py
Design Principles
No LLM at runtime
Predictable outcomes
Structured psychological reflection
Human-designed intelligence
Author
Nikitha Yadav
---
# Save File
```text id="f0c62s"
Ctrl + S