Principle:Googleapis Python genai History Management
| Knowledge Sources | |
|---|---|
| Domains | NLP, Conversational_AI |
| Last Updated | 2026-02-15 00:00 GMT |
Overview
A mechanism for retrieving and inspecting the accumulated conversation history within a chat session for logging, debugging, or session persistence.
Description
History Management provides access to the internal conversation log maintained by a chat session. The history is an ordered list of Content objects representing alternating user and model turns. A curated mode filters out turns that resulted in errors (e.g., safety blocks), providing a clean view of successful exchanges. History retrieval supports session persistence (saving/restoring conversations), debugging (inspecting what the model received), logging (audit trails), and display (showing conversation in a UI).
Usage
Use history management to inspect the conversation state, persist sessions across application restarts, display conversation logs, or debug unexpected model behavior. Use the curated=True parameter to get only successful turns, which is useful when restoring a session that had intermittent errors.
Theoretical Basis
History management implements the Event Log pattern:
# History as an append-only event log
history = []
# Each turn appends two events:
# 1. User message (role="user")
# 2. Model response (role="model")
# Curated view filters errors:
curated = [turn for turn in history if not turn.has_error]