8000 GitHub - genai-works-org/genai-protocol
[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Skip to content

genai-works-org/genai-protocol

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

23 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

GenAI Agent Protocol

GenAI Agent Protocol is an async‑first Python framework for building WebSocket‑based AI agents that let you:

  • Connect an agent to the GenAI.works ecosystem
  • Process messages via registered handler functions
  • Upload and retrieve files with contextual metadata (agent_context)
  • Log messages with contextual metadata (agent_context)

✨ Features

🧠 Agent Binding: Decorator-based agent registration
🪝 WebSocket Communication: Bidirectional messaging with a central server
📁 File Manager: Async file upload/download & metadata fetch
🪵 Context Logger: Structured, contextual WebSocket-based logging
🔎 OpenAI Schema Conversion: Automatically converts Pydantic-type function signatures to OpenAI-compatible schemas
📞 Agent‑to‑agent calls: invoke another registered agent from within your handler

📚 Core Concepts

GenAISession

A central controller that registers agents and manages the event lifecycle.

from genai_session.session import GenAISession

genai_session = GenAISession(
    jwt_token="<jwt received from GenAI CLI>",
    ws_url="<Router Container Public URL>", # Note if exposing infrastructure with ngrok
    api_base_url="<Backend Container Public URL>", # Note if exposing infrastructure with ngrok
)

@bind(...)

Registers a handler function with the session and make them visible to GenAI infrastructure.

from genai_session.session import GenAISession
from genai_session.utils.context import GenAIContext

genai_session = GenAISession(jwt_token="<jwt received from GenAI CLI>")

@genai_session.bind(name="Test Name", description="Test Description")
async def message_handler(agent_context: GenAIContext, parameter: str) -> str:
    ...

GenAIContext

Provides contextual info (agent_uuid, request_id, etc.), a logger, and access to the FileManager.

from genai_session.session import GenAISession
from genai_session.utils.context import GenAIContext

genai_session = GenAISession(jwt_token="<jwt received from GenAI CLI>")

@genai_session.bind(name="Test Name", description="Test Description")
async def message_handler(agent_context: GenAIContext, parameter: str) -> str:
    request_id = agent_context.request_id

Files

Handles file uploads (save) and retrievals (get_by_id, get_metadata_by_id).

from genai_session.session import GenAISession
from genai_session.utils.context import GenAIContext

genai_session = GenAISession(jwt_token="<jwt received from GenAI CLI>")

@genai_session.bind(name="txt_content_reader_agent", description="Agent returns txt file content")
async def get_file_content(agent_context: GenAIContext, file_id: str) -> str:
    file = await agent_context.files.get_by_id(file_id)
    file_metadata = await agent_context.files.get_metadata_by_id(file_id)
    ...

Logger

Sends JSON logs through WebSocket with severity levels (debug, info, warning, error, critical).

from genai_session.session import GenAISession
from genai_session.utils.context import GenAIContext

genai_session = GenAISession(jwt_token="<jwt received from GenAI CLI>")

@genai_session.bind()
async def reverse_name(agent_context: GenAIContext, name: str) -> str:
    """Agent reverses the name"""
    agent_context.logger.info("Inside the reverse_name function")
    agent_context.logger.debug(f"name: {name}")
    ...

Invoke Agent from Agent

You can invoke another agent from within an agent using the genai_session.send method.
This method takes the agent_uuid and params as arguments.

from genai_session.session import GenAISession
from genai_session.utils.context import GenAIContext
from genai_session.utils.agents import AgentResponse

genai_session = GenAISession(jwt_token="<jwt received from GenAI CLI>")

@genai_session.bind()
async def invoke_another_agent(agent_context: GenAIContext, name: dict) -> str:
    """Agent invokes another registered agent"""
    agent_response: AgentResponse = await genai_session.send(
        agent_uuid="agent_uuid", # you can get UUID from - await agent_context.get_my_agents()
        params={
            "username": name,
            "interests": ["python", "genai"],
            "age": 30,
        } # key is a parameter name, value is the value you want to pass
    )
    response = agent_response.response
    is_success = agent_response.is_success
    ...

📝 Function annotation examples

No parameters

@genai_session.bind(name="GetCurrentDate", description="Return current date")
async def get_current_date(agent_context: GenAIContext):
    ...

Built-in types

@genai_session.bind(name="Saver", description="Saves file")
async def file_saver(
    agent_context: GenAIContext,
    filename: str,
    file_content: str, 
    page_count: int, 
    images_names: list[str]
) -> dict:
    ...

Pydantic models

from pydantic import BaseModel, Field
from typing import List, Any


class TranslationInput(BaseModel):
    text: str = Field(..., description="Text to translate")
    language: str = Field(..., description="Code of the language to translate to (e.g. 'fr', 'es')")
    banned_words: List[str] = Field(..., description="List of words to be banned from translation")

@genai_session.bind(name="TranslationAgent", description="Translate the text into specified language")
async def get_translation(
    agent_context: GenAIContext,
    params: TranslationInput
) -> dict[str, Any]:
    text = params.text
    language = params.language
    banned_words = params.banned_words
    ...

typing Annotations

from typing import Any, Annotated

@genai_session.bind(name="TranslationAgent", description="Translate the text into specified language")
async def get_translation(
    agent_context: GenAIContext, 
    text: Annotated[str, "Text to translate"],
    language: Annotated[str, "Code of the language to translate to (e.g. 'fr', 'es')"],
    banned_words: Annotated[list[str], "List of words to be banned from translation"],
) -> dict[str, Any]:
    ...

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Contributors 2

  •  
  •  

Languages

0