rakam-systems-agent 0.1.1rc7__py3-none-any.whl

rakam_systems_agent-0.1.1rc7.dist-info/METADATA

@@ -0,0 +1,367 @@
+Metadata-Version: 2.4
+Name: rakam-systems-agent
+Version: 0.1.1rc7
+Summary: AI Agents framework with Pydantic AI support and LLM Gateway integration
+Project-URL: Homepage, https://github.com/Rakam-AI/rakam_systems-inhouse
+Project-URL: Documentation, https://github.com/Rakam-AI/rakam_systems-inhouse
+Project-URL: Repository, https://github.com/Rakam-AI/rakam_systems-inhouse
+Author-email: Mohamed Hilel <mohammedjassemhlel@gmail.com>, Peng Zheng <pengzheng990630@outlook.com>
+Classifier: License :: OSI Approved :: Apache Software License
+Classifier: Operating System :: OS Independent
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Requires-Python: >=3.10
+Requires-Dist: psycopg2-binary
+Requires-Dist: pydantic-ai<2.0.0,>=1.11.0
+Requires-Dist: pydantic>=2.11.5
+Requires-Dist: python-dotenv>=1.0.0
+Requires-Dist: pyyaml>=6.0
+Requires-Dist: rakam-system-core
+Provides-Extra: all
+Requires-Dist: rakam-system-agent[llm-providers]; extra == 'all'
+Provides-Extra: dev
+Requires-Dist: black>=23.0.0; extra == 'dev'
+Requires-Dist: pytest-asyncio>=0.21.0; extra == 'dev'
+Requires-Dist: pytest>=7.0.0; extra == 'dev'
+Requires-Dist: ruff>=0.1.0; extra == 'dev'
+Provides-Extra: llm-providers
+Requires-Dist: mistralai<2.0.0,>=1.9.0; extra == 'llm-providers'
+Requires-Dist: openai<3.0.0,>=1.37.0; extra == 'llm-providers'
+Requires-Dist: tiktoken; extra == 'llm-providers'
+Description-Content-Type: text/markdown
+
+# Rakam System Agent
+
+The agent package of Rakam Systems providing AI agent implementations powered by Pydantic AI.
+
+## Overview
+
+`rakam-system-agent` provides flexible AI agents with tool integration, chat history, and LLM gateway abstractions. This package depends on `rakam-system-core`.
+
+## Features
+
+- **Configuration-First Design**: Change agents without code changes - just update YAML files
+- **Async/Sync Support**: Full support for both synchronous and asynchronous agent operations
+- **Tool Integration**: Easy tool definition and integration using the `Tool.from_schema` pattern
+- **Model Settings**: Control model behavior including parallel tool calls, temperature, and max tokens
+- **Pydantic AI Powered**: Built on top of Pydantic AI library
+- **Streaming Support**: Both sync and async streaming interfaces
+- **Chat History**: Multiple backends (JSON, SQLite, PostgreSQL)
+- **LLM Gateway**: Unified interface for OpenAI and Mistral AI
+
+### 🎯 Configuration Convenience
+
+The agent package supports comprehensive YAML configuration, allowing you to:
+
+- **Switch LLM models** without changing code (GPT-4 → GPT-4o-mini → Claude)
+- **Tune parameters** instantly (temperature, max tokens, parallel tools)
+- **Modify prompts** without redeployment
+- **Add/remove tools** by editing config files
+- **Enable/disable tracking** via configuration
+- **Use different configs** for different environments
+
+**Example**: Switch from GPT-4o to GPT-4o-mini by changing one line in your YAML - no code changes, no redeployment needed!
+
+## Installation
+
+```bash
+# Requires core package
+pip install -e ./rakam-system-core
+
+# Install agent package
+pip install -e ./rakam-system-agent
+```
+
+## Quick Start
+
+### Using BaseAgent (Pydantic AI-powered)
+
+```python
+import asyncio
+from rakam_system_agent import BaseAgent
+from rakam_systems_core.ai_core.interfaces import ModelSettings
+from rakam_systems_core.ai_core.interfaces.tool import ToolComponent as Tool
+
+# Define a tool function
+async def get_weather(city: str) -> dict:
+    """Get weather information for a city"""
+    # Your implementation here
+    return {"city": city, "temperature": 72, "condition": "sunny"}
+
+# Create an agent with tools
+agent = BaseAgent(
+    name="weather_agent",
+    model="openai:gpt-4o",
+    system_prompt="You are a helpful weather assistant.",
+    tools=[
+        Tool.from_schema(
+            function=get_weather,
+            name='get_weather',
+            description='Get weather information for a city',
+            json_schema={
+                'type': 'object',
+                'properties': {
+                    'city': {'type': 'string', 'description': 'The city name'},
+                },
+                'required': ['city'],
+                'additionalProperties': False,
+            },
+            takes_ctx=False,
+        ),
+    ],
+)
+
+# Run the agent
+async def main():
+    result = await agent.arun(
+        "What's the weather in San Francisco?",
+        model_settings=ModelSettings(parallel_tool_calls=True),
+    )
+    print(result.output_text)
+
+asyncio.run(main())
+```
+
+## Core Components
+
+### AgentComponent
+
+The base abstract class for all agents. Provides:
+
+- `run()` / `arun()`: Execute the agent synchronously or asynchronously
+- `stream()` / `astream()`: Stream responses
+- Support for tools, model settings, and dependencies
+
+### BaseAgent
+
+The core agent implementation powered by Pydantic AI. This is the primary agent class in our system. Features:
+
+- Direct integration with Pydantic AI's Agent
+- Full support for parallel tool calls
+- Automatic conversion between our interfaces and Pydantic AI's
+- Support for both traditional tool lists and ToolRegistry/ToolInvoker system
+
+### Tool
+
+Wrapper for tool functions compatible with Pydantic AI's `Tool.from_schema` pattern:
+
+```python
+Tool.from_schema(
+    function=my_function,
+    name='my_function',
+    description='What this function does',
+    json_schema={
+        'type': 'object',
+        'properties': {...},
+        'required': [...],
+    },
+    takes_ctx=False,
+)
+```
+
+### ModelSettings
+
+Configure model behavior:
+
+```python
+ModelSettings(
+    parallel_tool_calls=True,  # Enable parallel tool execution
+    temperature=0.7,           # Control randomness
+    max_tokens=1000,           # Limit response length
+)
+```
+
+## Advanced Usage
+
+### Parallel vs Sequential Tool Calls
+
+Control whether tools are called in parallel or sequentially:
+
+```python
+# Parallel (faster for independent tools)
+result = await agent.arun(
+    "Get weather for NYC and LA",
+    model_settings=ModelSettings(parallel_tool_calls=True),
+)
+
+# Sequential (for dependent operations)
+result = await agent.arun(
+    "Get weather for NYC and LA",
+    model_settings=ModelSettings(parallel_tool_calls=False),
+)
+```
+
+### Using Dependencies
+
+Pass context/dependencies to your agent:
+
+```python
+class Deps:
+    def __init__(self, user_id: str):
+        self.user_id = user_id
+
+agent = BaseAgent(
+    deps_type=Deps,
+    # ...
+)
+
+result = await agent.arun(
+    "Process this",
+    deps=Deps(user_id="123"),
+)
+```
+
+### Streaming Responses
+
+```python
+async for chunk in agent.astream("Tell me a story"):
+    print(chunk, end='', flush=True)
+```
+
+## API Reference
+
+### AgentComponent
+
+```python
+class AgentComponent(BaseComponent):
+    def __init__(
+        self,
+        name: str,
+        config: Optional[Dict[str, Any]] = None,
+        model: Optional[str] = None,
+        deps_type: Optional[Type[Any]] = None,
+        system_prompt: Optional[str] = None,
+        tools: Optional[List[Any]] = None,
+    )
+
+    def run(
+        self,
+        input_data: Union[str, AgentInput],
+        deps: Optional[Any] = None,
+        model_settings: Optional[ModelSettings] = None
+    ) -> AgentOutput
+
+    async def arun(
+        self,
+        input_data: Union[str, AgentInput],
+        deps: Optional[Any] = None,
+        model_settings: Optional[ModelSettings] = None
+    ) -> AgentOutput
+```
+
+### Tool
+
+```python
+class Tool:
+    @classmethod
+    def from_schema(
+        cls,
+        function: Callable[..., Any],
+        name: str,
+        description: str,
+        json_schema: Dict[str, Any],
+        takes_ctx: bool = False,
+    ) -> "Tool"
+```
+
+### ModelSettings
+
+```python
+class ModelSettings:
+    def __init__(
+        self,
+        parallel_tool_calls: bool = True,
+        temperature: Optional[float] = None,
+        max_tokens: Optional[int] = None,
+        **kwargs: Any
+    )
+```
+
+## Examples
+
+See the `examples/ai_agents_examples/` directory in the main repository for complete examples demonstrating:
+
+- Multiple tool definitions
+- Parallel vs sequential tool calls
+- Performance comparisons
+- Complex multi-tool workflows
+- Chat history integration
+- RAG systems
+
+## Package Structure
+
+```
+rakam-system-agent/
+├── src/rakam_system_agent/
+│   ├── components/
+│   │   ├── base_agent.py         # BaseAgent (Pydantic AI-powered)
+│   │   ├── llm_gateway/          # LLM provider gateways
+│   │   ├── chat_history/         # Chat history backends
+│   │   ├── tools/                # Built-in tools
+│   │   └── __init__.py           # Exports
+│   └── server/                   # MCP server
+└── pyproject.toml
+```
+
+## Best Practices
+
+1. **Use async when possible**: Async operations are more efficient, especially with tools
+2. **Enable parallel tool calls**: For independent operations, parallel execution is much faster
+3. **Provide clear tool descriptions**: Better descriptions help the LLM use tools correctly
+4. **Use type hints**: JSON schemas should match your function signatures
+5. **Handle errors gracefully**: Tools should catch and return meaningful errors
+
+## Migration Guide
+
+### Using BaseAgent
+
+`BaseAgent` is the core Pydantic AI-powered agent implementation in this framework.
+
+```python
+from rakam_system_agent import BaseAgent
+
+agent = BaseAgent(
+    name="agent",
+    model="openai:gpt-4o",
+    system_prompt="You are a helpful assistant."
+)
+```
+
+## Troubleshooting
+
+### ImportError: pydantic_ai not installed
+
+Install Pydantic AI:
+
+```bash
+pip install pydantic_ai
+```
+
+### Tool not being called
+
+Check:
+
+1. Tool description is clear and relevant
+2. JSON schema matches function signature
+3. System prompt doesn't contradict tool usage
+
+### Performance issues
+
+- Enable `parallel_tool_calls=True` for independent operations
+- Use async functions for I/O-bound operations
+- Consider caching tool results when appropriate
+
+## Contributing
+
+When adding new agent types:
+
+1. Inherit from `BaseAgent`
+2. Implement `ainfer()` for async or `infer()` for sync
+3. Add tests in `tests/`
+4. Update this README with examples
+
+## License
+
+See main project LICENSE file.

rakam_systems_agent-0.1.1rc7.dist-info/RECORD

@@ -0,0 +1,23 @@
+rakam_systems_agent/__init__.py,sha256=SdBoUWf25RuwlcrJrNOlwMubRKi1CBAr0jjaRNLbnYo,586
+rakam_systems_agent/components/__init__.py,sha256=TmxsMqMo6I-VY18EngYqNz_iuUuXKQhJpry7x6wxWIc,525
+rakam_systems_agent/components/base_agent.py,sha256=oNEH0a8f1wxe2iK7V9rlGFKp5crEcFqx0cRsIwzv5ZY,13366
+rakam_systems_agent/components/chat_history/__init__.py,sha256=aMFn82ZxU6Y0R3WBbQIlgG7o03dhUEWrh27vSkrl8aI,319
+rakam_systems_agent/components/chat_history/json_chat_history.py,sha256=ILn8zMFA8NlUM9dcy02OBk5HXQYlQ6C8yOSF_zILexc,12379
+rakam_systems_agent/components/chat_history/postgres_chat_history.py,sha256=rl08LuTJbmC1NWSP1OAmo3RMYsrQsFSkFGupi7P_QCE,22718
+rakam_systems_agent/components/chat_history/sql_chat_history.py,sha256=8tYeWNQcYZANKZaDcYClEDtSb7GFW92xJG05lKyQ_iY,14853
+rakam_systems_agent/components/llm_gateway/README.md,sha256=wkqXBm-yKJiFzwOPGCsYbXbN8zLbhIN3zEikG6RhmjY,12815
+rakam_systems_agent/components/llm_gateway/__init__.py,sha256=OuFA26nseHHlbmWc-TjBo6nsiPWX9xty1r21it1BFL8,489
+rakam_systems_agent/components/llm_gateway/gateway_factory.py,sha256=M_paM8J3Y18svUXUVJclEGAevuuwwMmLSzG0j8iATAk,9821
+rakam_systems_agent/components/llm_gateway/mistral_gateway.py,sha256=1lgnNZ27CR0R1r4kSBS6aSqk-dpvjwcPAp4cscrrvV8,9010
+rakam_systems_agent/components/llm_gateway/openai_gateway.py,sha256=MgWLEZvCOKHOwPAhgDF8QhckFNdrUHkYbIGXLKfHPII,9283
+rakam_systems_agent/components/tools/LLM_GATEWAY_TOOLS_README.md,sha256=R7u1u_gXQc6gpnMyI_Z41JzeHz96Kj4GB-5gLBRuvMI,13514
+rakam_systems_agent/components/tools/__init__.py,sha256=ot215bK8aN-FmrDGLCAOa4Y7VvZ-wNEvw0AL-cvHGgQ,1058
+rakam_systems_agent/components/tools/example_tools.py,sha256=JMbXt8iMf3USB3ZeLEWP0S4b-oiDXW6kKpZl1RJe81o,13447
+rakam_systems_agent/components/tools/llm_gateway_tools.py,sha256=jSqfB_dd0iKQSLgzydLXgUeCQdYG3ZYVQ4GHIAHQ8VU,20669
+rakam_systems_agent/components/tools/search_tool.py,sha256=VpyyBgJiqHOAyljeztCFh7kwrCz7QuuNezyUXqx4vVo,497
+rakam_systems_agent/server/README.md,sha256=P3W_LY2cQPUyUC0gBHIO4RcwEs9b_UJqPzOzncHVL2w,10959
+rakam_systems_agent/server/__init__.py,sha256=Z1brWmrel319nrevleQCMqFumZrF2w1AmR-N9Cks4O4,211
+rakam_systems_agent/server/mcp_server_agent.py,sha256=bUO4ryRHmKA2qjlGVMtVbAmsCxw60wq6d0SJ0amP9BM,4346
+rakam_systems_agent-0.1.1rc7.dist-info/METADATA,sha256=vYTeV3Flz_TsSl2pxmTSapqHT4CisJIvkFg_u4cL3jU,10251
+rakam_systems_agent-0.1.1rc7.dist-info/WHEEL,sha256=WLgqFyCfm_KASv4WHyYy0P3pM_m7J5L9k2skdKLirC8,87
+rakam_systems_agent-0.1.1rc7.dist-info/RECORD,,

rakam_systems_agent-0.1.1rc7.dist-info/WHEEL

@@ -0,0 +1,4 @@
+Wheel-Version: 1.0
+Generator: hatchling 1.28.0
+Root-Is-Purelib: true
+Tag: py3-none-any