fix(DATAGO-128008):- enable auto-compaction out the gate & provide docs (#1279)

Author: enavitan

docs/docs/documentation/installing-and-configuring/session-storage.md

@@ -303,6 +303,22 @@ Parameters:
 - `database_url`: Connection string for the agent's database
 - `default_behavior`: `"PERSISTENT"` (reuse sessions) or `"RUN_BASED"` (new session per run)
 
+### Auto-Compaction for Long Conversations
+
+Agents automatically compact conversation history when context limits are exceeded. This feature is enabled by default and requires no configuration.
+
+When an LLM returns a context window overflow error, the system intercepts it and automatically compacts the first N% of conversation turns (where N is the compaction_percentage), then retries the request. You can control what percentage of history is compacted using the `compaction_percentage` setting:
+
+```yaml
+auto_summarization:
+  # Compact 50% of conversation history when limits are reached
+  # Default: 0.25 (25%)
+  # Range: 0.0 - 1.0 (e.g., 0.25 = 25%, 0.5 = 50%, 0.8 = 80%)
+  compaction_percentage: 0.5
+```
+
+The system always compacts up to the nearest complete conversation turn and preserves at least one recent turn uncompacted.
+
 ### Environment Variables
 
 Each agent can have its own database credentials:

examples/shared_config.yaml

@@ -144,10 +144,6 @@ shared_config:
 
     # default auto_summarization settings.
     auto_summarization: &default_auto_summarization
-      # Enable/disable automatic conversation history summarization
-      # Set to true to enable, false to disable (default: false)
-      enabled: ${SAM_ENABLE_AUTO_SUMMARIZATION, false}
-
       # Percentage of conversation to compact when summarisation is triggered
       # Conservative default: 0.25 = 25%
       # Range: 0.0 - 1.0 (e.g., 0.25 = 25%, 0.5 = 50%, 0.8 = 80%)

src/solace_agent_mesh/agent/adk/runner.py

@@ -1408,7 +1408,7 @@ async def run_adk_async_task_thread_wrapper(
     """
     # Read auto-summarization config from component (per-agent configuration)
     auto_sum_config = component.auto_summarization_config
-    compaction_enabled = auto_sum_config.get("enabled", False)
+    compaction_enabled = auto_sum_config.get("enabled", True)
     compaction_percentage = auto_sum_config.get("compaction_percentage", 0.25)
 
     logical_task_id = a2a_context.get("logical_task_id", "unknown_task")

src/solace_agent_mesh/agent/adk/services.py

@@ -519,7 +519,7 @@ def initialize_session_service(component) -> BaseSessionService:
 
     # Check if auto-summarization is enabled from component config
     auto_sum_config = component.auto_summarization_config
-    if auto_sum_config.get("enabled", False):
+    if auto_sum_config.get("enabled", True):
         # Wrap with FilteringSessionService to automatically filter ghost events
         # This ensures ALL get_session() calls across the codebase get filtered sessions
         # There is a risk of spilling summary events in case if flag flips between True/False - and no filtering.

src/solace_agent_mesh/agent/sac/component.py

@@ -206,7 +206,7 @@ def __init__(self, **kwargs):
             )
             self.auto_summarization_config = self.get_config(
                 "auto_summarization", {
-                    "enabled": False,
+                    "enabled": True,
                     "compaction_percentage": 0.25
                 }
             )

src/solace_agent_mesh/workflow/component.py

@@ -92,7 +92,7 @@ def __init__(self, **kwargs):
         workflow_config = self.get_config("workflow")
         self.auto_summarization_config = self.get_config(
             "auto_summarization", {
-                "enabled": False,
+                "enabled": True,
                 "compaction_percentage": 0.25
             }
         )

templates/shared_config.yaml

@@ -90,12 +90,8 @@ shared_config:
       max_result_preview_bytes: 4096
 
     # Auto-summarization configuration (prevents "too many tokens" errors)
-    # DISABLED by default - agents must explicitly enable it
+    # Enabled by default to prevent context limit issues in long conversations
     auto_summarization: &default_auto_summarization
-      # Enable/disable automatic conversation history summarization
-      # Set to true to enable, false to disable (default: false)
-      enabled: ${SAM_ENABLE_AUTO_SUMMARIZATION, false}
-
       # Percentage of conversation to compact when threshold exceeded
       # Conservative default: 0.25 = 25%
       # Range: 0.0 - 1.0 (e.g., 0.25 = 25%, 0.5 = 50%, 0.8 = 80%)