Skip to main content

​
ContextConfig

Defined in the models module.
AI Agent Complete context management configuration. Merges settings from CLI flags, env vars, and config files.

​
Properties

​
auto_compact
bool
No description available.
​
compact_threshold
float
No description available.
​
strategy
OptimizerStrategy
No description available.
​
output_reserve
int
No description available.
​
history_ratio
float
No description available.
​
tool_output_max
int
No description available.
​
default_tool_output_max
int
No description available.
​
prune_after_tokens
int
No description available.
​
protected_tools
List
No description available.
​
tool_limits
Dict
No description available.
​
monitor
MonitorConfig
No description available.
​
keep_recent_turns
int
No description available.
​
llm_summarize
bool
No description available.
​
smart_tool_summarize
bool
No description available.
​
session_tracking
bool
No description available.
​
track_summary
bool
No description available.
​
track_goal
bool
No description available.
​
track_plan
bool
No description available.
​
track_progress
bool
No description available.
​
aggregate_memory
bool
No description available.
​
aggregate_sources
List
No description available.
​
aggregate_max_tokens
int
No description available.
​
compression_min_gain_pct
float
No description available.
​
compression_max_attempts
int
No description available.
​
tool_budgets
Dict
No description available.
​
tool_summarize_limits
Dict
No description available.
​
estimation_mode
str
No description available.
​
log_estimation_mismatch
bool
No description available.
​
mismatch_threshold_pct
float
No description available.
​
monitor_enabled
bool
No description available.
​
monitor_path
str
No description available.
​
monitor_format
str
No description available.
​
monitor_frequency
str
No description available.
​
monitor_write_mode
str
No description available.
​
redact_sensitive
bool
No description available.
​
snapshot_timing
str
No description available.
​
allow_absolute_paths
bool
No description available.
​
source
str
No description available.

​
Methods

for_recipe()

Preset for recipe/workflow use cases with many tool calls.

from_env()

Load config from environment variables.

merge()

Create new config with overrides applied.
  • to_dict: Convert to dictionary.

​
Usage

# Enable with defaults
    agent = Agent(instructions="...", context=True)
    
    # Custom configuration
    agent = Agent(
        instructions="...",
        context=ContextConfig(
            auto_compact=True,
            session_tracking=True,      # Track goal/plan/progress
            aggregate_memory=True,       # Concurrent multi-memory fetch
        )
    )

​
Source

View on GitHub

praisonaiagents/context/models.py at line 246

Context Concept

Context Management

Context Strategies

Configuration Overview

Agent Config