Agentic core - Langgraph integration - Agent State Schema

[tkpratardan]

Agent state must track:

• Conversation history
• Dataset understanding (columns, types, issues)
• Generated cells and their results
• Analysis phase and progress
• Error tracking for self-correction

Scope

1. Define core state schema:
   It is a class with with state definition by; conversation, dataset context, cell tracking, analysis progress and process control
2. Define supporting schemas:
   - DataFrameInfo: columns, dtypes, shape, sample values, null counts
   - NotebookCell: id, type, content, result, error
   - ColumnInfo: name, dtype, statistics, detected issues
3. Implement state serialization for persistence

Expected outcomes

• State schema definitions
• Supporting model definitions
• State serialization utilities
• State migration strategy for schema changes

Requirements

• State captures all necessary information
• Annotated fields accumulate correctly
• State serializable for persistence
• Schema documented with examples

FYI @samarth9008 @Vedanshu7 @gpsaggese