Skip to main content

​
compaction

AI Agent Auto Context Compaction Module for PraisonAI Agents. Provides automatic context management for long conversations:
  • Anti-injection framing (prevents instruction hijacking)
  • Structured summary templates with organized sections
  • Iterative updates across multiple compactions
  • Message summarization with LLM integration
  • Context window management
  • Token counting and limits
  • Intelligent message pruning
Zero Performance Impact:
  • All imports are lazy loaded via getattr
  • Compaction only runs when needed
  • No overhead when context is within limits
Usage: from praisonaiagents import Agent, CompactionConfig from praisonaiagents.compaction import COMPACTION_PREFIX, SUMMARY_TEMPLATE

​
Agent-centric usage (recommended)

agent = Agent( name=“Assistant”, context=True # Anti-injection + structured template enabled )

​
With custom configuration

agent = Agent( name=“Assistant”, context=CompactionConfig( structured_template=True, compaction_prefix=“[REFERENCE ONLY] Use as background…”, iterative_update=True ) )

​
Access module constants

print(COMPACTION_PREFIX) # Default anti-injection prefix print(SUMMARY_TEMPLATE) # Structured template with placeholders

​
Import

from praisonaiagents import compaction