penguiflow 2.2.5__tar.gz → 2.2.6__tar.gz

{penguiflow-2.2.5/penguiflow.egg-info → penguiflow-2.2.6}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: penguiflow
-Version: 2.2.5
+Version: 2.2.6
 Summary: Async agent orchestration primitives.
 Author: PenguiFlow Team
 License: MIT License
@@ -44,6 +44,7 @@ Provides-Extra: a2a-server
 Requires-Dist: fastapi>=0.118; extra == "a2a-server"
 Provides-Extra: planner
 Requires-Dist: litellm>=1.77.3; extra == "planner"
+Requires-Dist: dspy>=3.0.3; extra == "planner"
 Dynamic: license-file
 
 # PenguiFlow 🐧❄️

penguiflow-2.2.6/examples/planner_enterprise_agent/__init__.py

@@ -0,0 +1,30 @@
+"""Enterprise agent example with ReactPlanner.
+
+This package provides a gold-standard implementation for production
+agent deployments using PenguiFlow's ReactPlanner with comprehensive
+observability, error handling, and configuration management.
+
+Public API:
+    - EnterpriseAgentOrchestrator: Main agent orchestrator
+    - AgentConfig: Type-safe configuration
+    - AgentTelemetry: Observability middleware
+
+Example:
+    >>> from examples.planner_enterprise_agent import (
+    ...     EnterpriseAgentOrchestrator,
+    ...     AgentConfig,
+    ... )
+    >>> config = AgentConfig.from_env()
+    >>> agent = EnterpriseAgentOrchestrator(config)
+    >>> result = await agent.execute("Analyze deployment logs")
+"""
+
+from .config import AgentConfig
+from .main import EnterpriseAgentOrchestrator
+from .telemetry import AgentTelemetry
+
+__all__ = [
+    "AgentConfig",
+    "AgentTelemetry",
+    "EnterpriseAgentOrchestrator",
+]

penguiflow-2.2.6/examples/planner_enterprise_agent/config.py

@@ -0,0 +1,93 @@
+"""Enterprise configuration management with environment variable support."""
+
+from __future__ import annotations
+
+import os
+from dataclasses import dataclass
+from typing import Literal
+
+
+@dataclass(frozen=True)
+class AgentConfig:
+    """Production-grade agent configuration.
+
+    All settings can be configured via environment variables with sensible
+    defaults for development. In production, set all credentials explicitly.
+    """
+
+    # LLM Configuration
+    llm_model: str
+    llm_temperature: float
+    llm_max_retries: int
+    llm_timeout_s: float
+
+    # Planner Configuration
+    planner_max_iters: int
+    planner_token_budget: int
+    planner_deadline_s: float | None
+    planner_hop_budget: int | None
+    planner_absolute_max_parallel: int
+
+    # Summarizer LLM (cheaper model for trajectory compression)
+    summarizer_model: str | None
+
+    # Observability
+    log_level: Literal["DEBUG", "INFO", "WARNING", "ERROR"]
+    enable_telemetry: bool
+    telemetry_backend: Literal["logging", "mlflow", "datadog"]
+    mlflow_tracking_uri: str | None
+
+    # Application Settings
+    environment: Literal["development", "staging", "production"]
+    agent_name: str
+
+    @classmethod
+    def from_env(cls) -> AgentConfig:
+        """Load configuration from environment variables.
+
+        Required environment variables:
+        - OPENAI_API_KEY or other LLM provider credentials
+
+        Optional (with defaults):
+        - LLM_MODEL (default: gpt-4o-mini)
+        - AGENT_ENVIRONMENT (default: development)
+        - ENABLE_TELEMETRY (default: true)
+        - LOG_LEVEL (default: INFO)
+        """
+        return cls(
+            # LLM settings
+            llm_model=os.getenv("LLM_MODEL", "gpt-4o-mini"),
+            llm_temperature=float(os.getenv("LLM_TEMPERATURE", "0.0")),
+            llm_max_retries=int(os.getenv("LLM_MAX_RETRIES", "3")),
+            llm_timeout_s=float(os.getenv("LLM_TIMEOUT_S", "60.0")),
+            # Planner settings
+            planner_max_iters=int(os.getenv("PLANNER_MAX_ITERS", "12")),
+            planner_token_budget=int(os.getenv("PLANNER_TOKEN_BUDGET", "8000")),
+            planner_deadline_s=_parse_optional_float(
+                os.getenv("PLANNER_DEADLINE_S")
+            ),
+            planner_hop_budget=_parse_optional_int(os.getenv("PLANNER_HOP_BUDGET")),
+            planner_absolute_max_parallel=int(
+                os.getenv("PLANNER_ABSOLUTE_MAX_PARALLEL", "50")
+            ),
+            # Summarizer
+            summarizer_model=os.getenv("SUMMARIZER_MODEL", "gpt-4o-mini"),
+            # Observability
+            log_level=os.getenv("LOG_LEVEL", "INFO"),  # type: ignore
+            enable_telemetry=os.getenv("ENABLE_TELEMETRY", "true").lower() == "true",
+            telemetry_backend=os.getenv("TELEMETRY_BACKEND", "logging"),  # type: ignore
+            mlflow_tracking_uri=os.getenv("MLFLOW_TRACKING_URI"),
+            # Application
+            environment=os.getenv("AGENT_ENVIRONMENT", "development"),  # type: ignore
+            agent_name=os.getenv("AGENT_NAME", "enterprise_agent"),
+        )
+
+
+def _parse_optional_float(value: str | None) -> float | None:
+    """Parse optional float from environment variable."""
+    return float(value) if value is not None else None
+
+
+def _parse_optional_int(value: str | None) -> int | None:
+    """Parse optional int from environment variable."""
+    return int(value) if value is not None else None