Release v4.3.1

Author: MervinPraison

docker/Dockerfile.chat

@@ -16,7 +16,7 @@ RUN mkdir -p /root/.praison
 # Install Python packages (using latest versions)
 RUN pip install --no-cache-dir \
     praisonai_tools \
-    "praisonai>=4.3.0" \
+    "praisonai>=4.3.1" \
     "praisonai[chat]" \
     "embedchain[github,youtube]"
 

docker/Dockerfile.dev

@@ -20,7 +20,7 @@ RUN mkdir -p /root/.praison
 # Install Python packages (using latest versions)
 RUN pip install --no-cache-dir \
     praisonai_tools \
-    "praisonai>=4.3.0" \
+    "praisonai>=4.3.1" \
     "praisonai[ui]" \
     "praisonai[chat]" \
     "praisonai[realtime]" \

docker/Dockerfile.ui

@@ -16,7 +16,7 @@ RUN mkdir -p /root/.praison
 # Install Python packages (using latest versions)
 RUN pip install --no-cache-dir \
     praisonai_tools \
-    "praisonai>=4.3.0" \
+    "praisonai>=4.3.1" \
     "praisonai[ui]" \
     "praisonai[crewai]"
 

src/praisonai-agents/examples/streaming_reasoning_example.py

@@ -0,0 +1,175 @@
+"""
+Streaming with Reasoning Content Example
+
+This example demonstrates how to handle streaming events with reasoning content
+from models like o1, o1-mini, and deepseek-reasoner.
+
+The is_reasoning flag allows you to distinguish between:
+- Reasoning/thinking content (is_reasoning=True)
+- Final response content (is_reasoning=False)
+"""
+
+from praisonaiagents import Agent
+from praisonaiagents.streaming import StreamEvent, StreamEventType, StreamMetrics
+
+
+def create_styled_callback():
+    """Create a callback that styles reasoning content differently."""
+    
+    def callback(event: StreamEvent):
+        if event.type == StreamEventType.DELTA_TEXT:
+            if event.is_reasoning:
+                # Dim gray for reasoning content
+                print(f"\033[90m{event.content}\033[0m", end="", flush=True)
+            else:
+                # Normal for response content
+                print(event.content, end="", flush=True)
+        elif event.type == StreamEventType.STREAM_END:
+            print()  # Newline at end
+    
+    return callback
+
+
+def create_collecting_callback():
+    """Create a callback that collects reasoning and response separately."""
+    reasoning_parts = []
+    response_parts = []
+    
+    def callback(event: StreamEvent):
+        if event.type == StreamEventType.DELTA_TEXT:
+            if event.is_reasoning:
+                reasoning_parts.append(event.content)
+            else:
+                response_parts.append(event.content)
+    
+    def get_results():
+        return {
+            "reasoning": "".join(reasoning_parts),
+            "response": "".join(response_parts)
+        }
+    
+    return callback, get_results
+
+
+def example_basic_streaming():
+    """Basic streaming with reasoning detection."""
+    print("=" * 60)
+    print("Example 1: Basic Streaming with Reasoning Detection")
+    print("=" * 60)
+    
+    agent = Agent(
+        name="thinker",
+        instructions="Think step by step before answering.",
+        llm="gpt-4o-mini"  # Change to o1-mini for actual reasoning
+    )
+    
+    # Add styled callback
+    agent.stream_emitter.add_callback(create_styled_callback())
+    
+    # Enable metrics
+    agent.stream_emitter.enable_metrics()
+    
+    # Run with streaming
+    response = agent.start("What is 15 * 23?", stream=True)
+    
+    # Show metrics
+    metrics = agent.stream_emitter.get_metrics()
+    if metrics:
+        print(f"\n📊 {metrics.format_summary()}")
+    
+    return response
+
+
+def example_collect_separately():
+    """Collect reasoning and response content separately."""
+    print("\n" + "=" * 60)
+    print("Example 2: Collect Reasoning and Response Separately")
+    print("=" * 60)
+    
+    agent = Agent(
+        name="analyzer",
+        instructions="Analyze the problem carefully.",
+        llm="gpt-4o-mini"
+    )
+    
+    # Create collecting callback
+    callback, get_results = create_collecting_callback()
+    agent.stream_emitter.add_callback(callback)
+    
+    # Run with streaming
+    response = agent.start("Explain why the sky is blue.", stream=True)
+    
+    # Get separated results
+    results = get_results()
+    
+    print(f"\n📝 Reasoning length: {len(results['reasoning'])} chars")
+    print(f"📝 Response length: {len(results['response'])} chars")
+    
+    return response
+
+
+def example_with_metrics():
+    """Track streaming performance metrics."""
+    print("\n" + "=" * 60)
+    print("Example 3: Streaming with Performance Metrics")
+    print("=" * 60)
+    
+    metrics = StreamMetrics()
+    
+    def metrics_callback(event: StreamEvent):
+        metrics.update_from_event(event)
+        
+        if event.type == StreamEventType.FIRST_TOKEN:
+            print("⚡ First token received!")
+        elif event.type == StreamEventType.DELTA_TEXT:
+            print(event.content, end="", flush=True)
+        elif event.type == StreamEventType.STREAM_END:
+            print()
+    
+    agent = Agent(name="assistant", llm="gpt-4o-mini")
+    agent.stream_emitter.add_callback(metrics_callback)
+    
+    response = agent.start("Write a haiku about coding.", stream=True)
+    
+    print("\n📊 Performance Metrics:")
+    print(f"   TTFT: {metrics.ttft * 1000:.0f}ms")
+    print(f"   Stream Duration: {metrics.stream_duration * 1000:.0f}ms")
+    print(f"   Total Time: {metrics.total_time * 1000:.0f}ms")
+    print(f"   Tokens: {metrics.token_count}")
+    print(f"   Speed: {metrics.tokens_per_second:.1f} tokens/sec")
+    
+    return response
+
+
+def example_multi_agent_context():
+    """Show agent context in multi-agent scenarios."""
+    print("\n" + "=" * 60)
+    print("Example 4: Multi-Agent Context")
+    print("=" * 60)
+    
+    def context_callback(event: StreamEvent):
+        if event.type == StreamEventType.DELTA_TEXT:
+            prefix = f"[{event.agent_id}] " if event.agent_id else ""
+            print(f"{prefix}{event.content}", end="", flush=True)
+        elif event.type == StreamEventType.STREAM_END:
+            print()
+    
+    # Note: agent_id is populated when using multi-agent workflows
+    agent = Agent(name="helper", llm="gpt-4o-mini")
+    agent.stream_emitter.add_callback(context_callback)
+    
+    response = agent.start("Say hello!", stream=True)
+    
+    return response
+
+
+if __name__ == "__main__":
+    print("🚀 PraisonAI Streaming with Reasoning Examples\n")
+    
+    # Run examples
+    example_basic_streaming()
+    example_collect_separately()
+    example_with_metrics()
+    example_multi_agent_context()
+    
+    print("\n✅ All examples completed!")

src/praisonai-agents/praisonaiagents/agent/agent.py

@@ -150,6 +150,7 @@ def _is_file_path(value: str) -> bool:
     from ..task.task import Task
     from .handoff import Handoff, HandoffConfig, HandoffResult
     from ..rag.models import RAGResult, ContextPack
+    from ..eval.results import EvaluationLoopResult
 
 class Agent:
     # Class-level counter for generating unique display names for nameless agents
@@ -851,6 +852,24 @@ def __init__(
         elif memory is False:
             memory = None
 
+        # ─────────────────────────────────────────────────────────────────────
+        # Resolve HISTORY from MemoryConfig - FAST PATH
+        # History is enabled via memory="history" preset or MemoryConfig(history=True)
+        # ─────────────────────────────────────────────────────────────────────
+        _history_enabled = False
+        _history_limit = 10
+        _history_session_id = None
+        
+        # Check if memory config has history settings
+        if _memory_config is not None and isinstance(_memory_config, MemoryConfig):
+            if _memory_config.history:
+                _history_enabled = True
+                _history_limit = _memory_config.history_limit
+        
+        # Use auto_save session if no explicit session and history is enabled
+        if _history_enabled and _history_session_id is None and auto_save:
+            _history_session_id = auto_save
+        
         # ─────────────────────────────────────────────────────────────────────
         # Resolve KNOWLEDGE param - FAST PATH
         # ─────────────────────────────────────────────────────────────────────
@@ -1353,6 +1372,11 @@ def __init__(
         # Used when session_id is provided but no DB adapter
         self._session_store = None
         self._session_store_initialized = False
+        
+        # History injection settings (for auto-injecting session history into context)
+        self._history_enabled = _history_enabled
+        self._history_limit = _history_limit
+        self._history_session_id = _history_session_id
 
         # Agent-centric feature instances (lazy loaded for zero performance impact)
         self._auto_memory = auto_memory
@@ -2298,6 +2322,99 @@ def handoff_to(
         )
         return handoff_obj.execute_programmatic(self, prompt, context)
 
+    def run_until(
+        self,
+        prompt: str,
+        criteria: str,
+        threshold: float = 8.0,
+        max_iterations: int = 5,
+        mode: str = "optimize",
+        on_iteration: Optional[Callable[[Any], None]] = None,
+        verbose: bool = False,
+    ) -> "EvaluationLoopResult":
+        """
+        Run agent iteratively until output meets quality criteria.
+        
+        This method implements the "Ralph Loop" pattern: run agent → judge output
+        → improve based on feedback → repeat until threshold met.
+        
+        Args:
+            prompt: The prompt to send to the agent
+            criteria: Evaluation criteria for the Judge (e.g., "Response is thorough")
+            threshold: Score threshold for success (default: 8.0, scale 1-10)
+            max_iterations: Maximum iterations before stopping (default: 5)
+            mode: "optimize" (stop on success) or "review" (run all iterations)
+            on_iteration: Optional callback called after each iteration
+            verbose: Enable verbose logging
+            
+        Returns:
+            EvaluationLoopResult with iteration history and final score
+            
+        Example:
+            ```python
+            agent = Agent(name="analyzer", instructions="Analyze systems")
+            result = agent.run_until(
+                "Analyze the auth flow",
+                criteria="Analysis is thorough and actionable",
+                threshold=8.0,
+            )
+            print(result.final_score)  # 8.5
+            print(result.success)      # True
+            ```
+        """
+        from ..eval.loop import EvaluationLoop
+        
+        loop = EvaluationLoop(
+            agent=self,
+            criteria=criteria,
+            threshold=threshold,
+            max_iterations=max_iterations,
+            mode=mode,
+            on_iteration=on_iteration,
+            verbose=verbose,
+        )
+        return loop.run(prompt)
+    
+    async def run_until_async(
+        self,
+        prompt: str,
+        criteria: str,
+        threshold: float = 8.0,
+        max_iterations: int = 5,
+        mode: str = "optimize",
+        on_iteration: Optional[Callable[[Any], None]] = None,
+        verbose: bool = False,
+    ) -> "EvaluationLoopResult":
+        """
+        Async version of run_until().
+        
+        Run agent iteratively until output meets quality criteria.
+        
+        Args:
+            prompt: The prompt to send to the agent
+            criteria: Evaluation criteria for the Judge
+            threshold: Score threshold for success (default: 8.0)
+            max_iterations: Maximum iterations before stopping (default: 5)
+            mode: "optimize" (stop on success) or "review" (run all iterations)
+            on_iteration: Optional callback called after each iteration
+            verbose: Enable verbose logging
+            
+        Returns:
+            EvaluationLoopResult with iteration history and final score
+        """
+        from ..eval.loop import EvaluationLoop
+        
+        loop = EvaluationLoop(
+            agent=self,
+            criteria=criteria,
+            threshold=threshold,
+            max_iterations=max_iterations,
+            mode=mode,
+            on_iteration=on_iteration,
+            verbose=verbose,
+        )
+        return await loop.run_async(prompt)
+    
     async def handoff_to_async(
         self,
         target_agent: 'Agent',
@@ -3442,7 +3559,19 @@ def _build_messages(self, prompt, temperature=1.0, output_json=None, output_pyda
             if system_prompt:
                 messages.append({"role": "system", "content": system_prompt})
 
-            # Add chat history
+            # Inject session history if enabled (from persistent storage)
+            if self._history_enabled and self._session_store is not None:
+                try:
+                    session_history = self._session_store.get_chat_history(
+                        self._history_session_id,
+                        max_messages=self._history_limit
+                    )
+                    if session_history:
+                        messages.extend(session_history)
+                except Exception as e:
+                    logging.debug(f"Failed to load session history: {e}")
+            
+            # Add in-memory chat history (current conversation)
             if self.chat_history:
                 messages.extend(self.chat_history)
 

src/praisonai-agents/praisonaiagents/config/feature_configs.py

@@ -165,6 +165,10 @@ class MemoryConfig:
     # Continuous learning configuration
     learn: Optional[Union[bool, LearnConfig]] = None
 
+    # History injection (auto-inject session history into context)
+    history: bool = False
+    history_limit: int = 10
+    
     def to_dict(self) -> Dict[str, Any]:
         """Convert to dictionary."""
         learn_dict = None
@@ -181,6 +185,8 @@ def to_dict(self) -> Dict[str, Any]:
             "claude_memory": self.claude_memory,
             "config": self.config,
             "learn": learn_dict,
+            "history": self.history,
+            "history_limit": self.history_limit,
         }
 
 

src/praisonai-agents/praisonaiagents/config/presets.py

@@ -26,6 +26,10 @@
     "mongodb": {"backend": "mongodb"},
     # Feature presets
     "learn": {"backend": "file", "learn": True},  # Enable Agent Learn with file backend
+    # History presets (auto-inject session history into context)
+    "history": {"backend": "file", "history": True, "history_limit": 10},
+    "session": {"backend": "file", "history": True, "history_limit": 10},  # Alias
+    "chat": {"backend": "file", "history": True, "history_limit": 20},  # Conversational
 }
 
 MEMORY_URL_SCHEMES: Dict[str, str] = {