versionhq 1.1.10.5__tar.gz → 1.1.10.7__tar.gz

{versionhq-1.1.10.5 → versionhq-1.1.10.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: versionhq
-Version: 1.1.10.5
+Version: 1.1.10.7
 Summary: LLM orchestration frameworks for model-agnostic AI agents that handle complex outbound workflows
 Author-email: Kuriko Iwai <kuriko@versi0n.io>
 License: MIT License
@@ -66,32 +66,17 @@ Requires-Dist: composio-langchain>=0.6.12
 ![pyenv ver](https://img.shields.io/badge/pyenv-2.5.0-orange)
 
 
-An LLM orchestration frameworks for multi-agent systems with RAG to autopilot outbound workflows.
-
-Agents are model agnostic.
-
-Messaging workflows are created at individual level, and will be deployed on third-party services via `Composio`.
-
+LLM orchestration frameworks to deploy multi-agent systems focusing on complex outbound tasks.
 
 **Visit:**
 
 - [PyPI](https://pypi.org/project/versionhq/)
-- [Github (LLM orchestration)](https://github.com/versionHQ/multi-agent-system)
-- [Doc](https://chief-oxygen-8a2.notion.site/Documentation-17e923685cf98001a5fad5c4b2acd79b?pvs=4)
-<!-- - [Github (Test client app)](https://github.com/versionHQ/test-client-app) -->
-- [Use case](https://versi0n.io/): Client web app for the outbound messaging automation
+- [Github (LLM orchestration framework)](https://github.com/versionHQ/multi-agent-system)
+- [Github (Test client app)](https://github.com/versionHQ/test-client-app)
+- [Use case](https://versi0n.io/playground) / [Quick demo](https://res.cloudinary.com/dfeirxlea/video/upload/v1737732977/pj_m_home/pnsyh5mfvmilwgt0eusa.mov)
+- [Documentation](https://chief-oxygen-8a2.notion.site/Documentation-17e923685cf98001a5fad5c4b2acd79b?pvs=4) *Some components are under review.
 
 
-<hr />
-
-## Mindmap
-
-LLM-powered `agent`s and `team`s use `tool`s and their own knowledge to complete the `task` given by the client or the system.
-
-<p align="center">
-   <img src="https://res.cloudinary.com/dfeirxlea/image/upload/v1733556715/pj_m_home/urwte15at3h0dr8mdlyo.png" alt="mindmap" width="400">
-</p>
-
 <hr />
 
 ## Table of Content
@@ -99,8 +84,10 @@ LLM-powered `agent`s and `team`s use `tool`s and their own knowledge to complete
 <!-- DON'T EDIT THIS SECTION, INSTEAD RE-RUN doctoc TO UPDATE -->
 
 - [Key Features](#key-features)
-- [Usage](#usage)
-  - [Case 1. Build an AI agent on LLM of your choice and execute a task:](#case-1-build-an-ai-agent-on-llm-of-your-choice-and-execute-a-task)
+  - [Agent formation](#agent-formation)
+- [Quick Start](#quick-start)
+  - [Case 1. Single agent network:](#case-1-single-agent-network)
+    - [Return a structured output with a summary in string.](#return-a-structured-output-with-a-summary-in-string)
   - [Case 2. Form a team to handle multiple tasks:](#case-2-form-a-team-to-handle-multiple-tasks)
 - [Technologies Used](#technologies-used)
 - [Project Structure](#project-structure)
@@ -119,68 +106,75 @@ LLM-powered `agent`s and `team`s use `tool`s and their own knowledge to complete
 
 ## Key Features
 
-A mulit-agent systems with Rag that tailors messaging workflow, predicts its performance, and deploys it on third-party tools.
+Generate mulit-agent systems depending on the complexity of the task, and execute the task with agents of choice.
+
+Model-agnostic agents can handle RAG tools, tools, callbacks, and knowledge sharing among other agents.
 
-The `agent` is model agnostic. The default model is set Chat GTP 4o. We ask the client their preference and switch it accordingly using llm variable stored in the `BaseAgent` class.
 
-Multiple `agents` can form a `team` to complete complex tasks together.
+###  Agent formation
+Depending on the task complexity, agents can make a different formation.
 
-**1. Analysis**
-- Professional `agents` handle the analysis `tasks` on each client, customer, and product.
+You can specify which formation you want them to generate, or let the agent decide if you don’t have a clear plan.
 
-**2. Messaging Workflow Creation**
-- Several `teams` receive the analysis and design initial messaging workflow with several layers.
-- Ask the client for their inputs
-- Deploy the workflow on the third party tools using `composio`.
 
-**3. Autopiloting**
-- Responsible `agents` or `teams` autopilot executing and refining the messaging workflow.
+|  | **Solo Agent** | **Supervising** | **Network** | **Random** |
+| :--- | :--- | :--- | :--- | :--- |
+| **Formation** | <img src="https://res.cloudinary.com/dfeirxlea/image/upload/v1737893140/pj_m_agents/tglrxoiuv7kk7nzvpe1z.jpg" alt="solo" width="200"> | <img src="https://res.cloudinary.com/dfeirxlea/image/upload/v1737893141/pj_m_agents/hxngdvnn5b5qdxo0ayl5.jpg" alt="solo" width="200"> | <img src="https://res.cloudinary.com/dfeirxlea/image/upload/v1737893142/pj_m_agents/kyc6neg8m6keyizszcpi.jpg" alt="solo" width="200"> | <img src="https://res.cloudinary.com/dfeirxlea/image/upload/v1737893139/pj_m_agents/hzpchlcpnpwxwaapu1hr.jpg" alt="solo" width="200"> |
+| **Usage** | <ul><li>A single agent with tools, knowledge, and memory.</li><li>When self-learning mode is on - it will turn into **Random** formation.</li></ul> | <ul><li>Leader agent gives directions, while sharing its knowledge and memory.</li><li>Subordinates can be solo agents or networks.</li></ul> | <ul><li>Share tasks, knowledge, and memory among network members.</li></ul> | <ul><li>A single agent handles tasks, asking help from other agents without sharing its memory or knowledge.</li></ul> |
+| **Use case** | An email agent drafts promo message for the given audience. | The leader agent strategizes an outbound campaign plan and assigns components such as media mix or message creation to subordinate agents. | An email agent and social media agent share the product knowledge and deploy multi-channel outbound campaign. | 1. An email agent drafts promo message for the given audience, asking insights on tones from other email agents which oversee other clusters. 2. An agent calls the external agent to deploy the campaign. |
 
 <hr />
 
-## Usage
+## Quick Start
+
+**Install `versionhq` package:**
 
-1. Install `versionhq` package:
    ```
-   uv pip install versionhq
+   pip install versionhq
    ```
 
-2. You can use the `versionhq` module in your Python app.
+(Python >= 3.13)
+
 
+### Case 1. Single agent network:
 
-### Case 1. Build an AI agent on LLM of your choice and execute a task:
+#### Return a structured output with a summary in string.
 
    ```
+   from pydantic import BaseModel
    from versionhq.agent.model import Agent
-   from versionhq.task.model import Task, ResponseField
+   from versionhq.task.model import Task
 
-   def my_callback_func():
-      """callback func"""
+   class CustomOutput(BaseModel):
+      test1: str
+      test2: list[str]
+
+   def dummy_func(message: str, test1: str, test2: list[str]) -> str:
+      return f"{message}: {test1}, {", ".join(test2)}"
 
-   agent = Agent(
-      role="demo",
-      goal="amazing project goal",
-      skillsets=["skill_1", "skill_2", ],
-      tools=["amazing RAG tool",]
-      llm="llm-of-your-choice"
-   )
+
+   agent = Agent(role="demo", goal="amazing project goal")
 
    task = Task(
       description="Amazing task",
-      response_fields=[
-         ResponseField(title="test1", data_type=str, required=True),
-         ResponseField(title="test2", data_type=list, items=str, required=True),
-      ],
-      callbacks=[my_callback_func]
+      pydantic_custom_output=CustomOutput,
+      callback=dummy_func,
+      callback_kwargs=dict(message="Hi! Here is the result: ")
    )
+
    res = task.execute_sync(agent=agent, context="amazing context to consider.")
-   return res.to_dict()
+   print(res)
    ```
 
-This will return a dictionary with keys defined in the `ResponseField`.
+This will return `TaskOutput` that stores a response in string, JSON dict, and Pydantic model: `CustomOutput` formats with a callback result.
 
    ```
-   { test1: "answer1", "test2": ["answer2-1", "answer2-2", "answer2-3",] }
+   res == TaskOutput(
+      raw="{\\"test1\\": \\"random str\\", \\"test2\\": [\\"item1\\", \\"item2\\"]}",
+      json_dict={"test1": "random str", "test2": ["item1", "item2"]},
+      pydantic=CustomOutput(test1="random str", test2=["item 1", "item 2"]),
+      callback_output="Hi! Here is the result: random str, item 1, item 2",
+   )
    ```
 
 ### Case 2. Form a team to handle multiple tasks:

{versionhq-1.1.10.5 → versionhq-1.1.10.7}/README.md

@@ -7,32 +7,17 @@
 ![pyenv ver](https://img.shields.io/badge/pyenv-2.5.0-orange)
 
 
-An LLM orchestration frameworks for multi-agent systems with RAG to autopilot outbound workflows.
-
-Agents are model agnostic.
-
-Messaging workflows are created at individual level, and will be deployed on third-party services via `Composio`.
-
+LLM orchestration frameworks to deploy multi-agent systems focusing on complex outbound tasks.
 
 **Visit:**
 
 - [PyPI](https://pypi.org/project/versionhq/)
-- [Github (LLM orchestration)](https://github.com/versionHQ/multi-agent-system)
-- [Doc](https://chief-oxygen-8a2.notion.site/Documentation-17e923685cf98001a5fad5c4b2acd79b?pvs=4)
-<!-- - [Github (Test client app)](https://github.com/versionHQ/test-client-app) -->
-- [Use case](https://versi0n.io/): Client web app for the outbound messaging automation
+- [Github (LLM orchestration framework)](https://github.com/versionHQ/multi-agent-system)
+- [Github (Test client app)](https://github.com/versionHQ/test-client-app)
+- [Use case](https://versi0n.io/playground) / [Quick demo](https://res.cloudinary.com/dfeirxlea/video/upload/v1737732977/pj_m_home/pnsyh5mfvmilwgt0eusa.mov)
+- [Documentation](https://chief-oxygen-8a2.notion.site/Documentation-17e923685cf98001a5fad5c4b2acd79b?pvs=4) *Some components are under review.
 
 
-<hr />
-
-## Mindmap
-
-LLM-powered `agent`s and `team`s use `tool`s and their own knowledge to complete the `task` given by the client or the system.
-
-<p align="center">
-   <img src="https://res.cloudinary.com/dfeirxlea/image/upload/v1733556715/pj_m_home/urwte15at3h0dr8mdlyo.png" alt="mindmap" width="400">
-</p>
-
 <hr />
 
 ## Table of Content
@@ -40,8 +25,10 @@ LLM-powered `agent`s and `team`s use `tool`s and their own knowledge to complete
 <!-- DON'T EDIT THIS SECTION, INSTEAD RE-RUN doctoc TO UPDATE -->
 
 - [Key Features](#key-features)
-- [Usage](#usage)
-  - [Case 1. Build an AI agent on LLM of your choice and execute a task:](#case-1-build-an-ai-agent-on-llm-of-your-choice-and-execute-a-task)
+  - [Agent formation](#agent-formation)
+- [Quick Start](#quick-start)
+  - [Case 1. Single agent network:](#case-1-single-agent-network)
+    - [Return a structured output with a summary in string.](#return-a-structured-output-with-a-summary-in-string)
   - [Case 2. Form a team to handle multiple tasks:](#case-2-form-a-team-to-handle-multiple-tasks)
 - [Technologies Used](#technologies-used)
 - [Project Structure](#project-structure)
@@ -60,68 +47,75 @@ LLM-powered `agent`s and `team`s use `tool`s and their own knowledge to complete
 
 ## Key Features
 
-A mulit-agent systems with Rag that tailors messaging workflow, predicts its performance, and deploys it on third-party tools.
+Generate mulit-agent systems depending on the complexity of the task, and execute the task with agents of choice.
+
+Model-agnostic agents can handle RAG tools, tools, callbacks, and knowledge sharing among other agents.
 
-The `agent` is model agnostic. The default model is set Chat GTP 4o. We ask the client their preference and switch it accordingly using llm variable stored in the `BaseAgent` class.
 
-Multiple `agents` can form a `team` to complete complex tasks together.
+###  Agent formation
+Depending on the task complexity, agents can make a different formation.
 
-**1. Analysis**
-- Professional `agents` handle the analysis `tasks` on each client, customer, and product.
+You can specify which formation you want them to generate, or let the agent decide if you don’t have a clear plan.
 
-**2. Messaging Workflow Creation**
-- Several `teams` receive the analysis and design initial messaging workflow with several layers.
-- Ask the client for their inputs
-- Deploy the workflow on the third party tools using `composio`.
 
-**3. Autopiloting**
-- Responsible `agents` or `teams` autopilot executing and refining the messaging workflow.
+|  | **Solo Agent** | **Supervising** | **Network** | **Random** |
+| :--- | :--- | :--- | :--- | :--- |
+| **Formation** | <img src="https://res.cloudinary.com/dfeirxlea/image/upload/v1737893140/pj_m_agents/tglrxoiuv7kk7nzvpe1z.jpg" alt="solo" width="200"> | <img src="https://res.cloudinary.com/dfeirxlea/image/upload/v1737893141/pj_m_agents/hxngdvnn5b5qdxo0ayl5.jpg" alt="solo" width="200"> | <img src="https://res.cloudinary.com/dfeirxlea/image/upload/v1737893142/pj_m_agents/kyc6neg8m6keyizszcpi.jpg" alt="solo" width="200"> | <img src="https://res.cloudinary.com/dfeirxlea/image/upload/v1737893139/pj_m_agents/hzpchlcpnpwxwaapu1hr.jpg" alt="solo" width="200"> |
+| **Usage** | <ul><li>A single agent with tools, knowledge, and memory.</li><li>When self-learning mode is on - it will turn into **Random** formation.</li></ul> | <ul><li>Leader agent gives directions, while sharing its knowledge and memory.</li><li>Subordinates can be solo agents or networks.</li></ul> | <ul><li>Share tasks, knowledge, and memory among network members.</li></ul> | <ul><li>A single agent handles tasks, asking help from other agents without sharing its memory or knowledge.</li></ul> |
+| **Use case** | An email agent drafts promo message for the given audience. | The leader agent strategizes an outbound campaign plan and assigns components such as media mix or message creation to subordinate agents. | An email agent and social media agent share the product knowledge and deploy multi-channel outbound campaign. | 1. An email agent drafts promo message for the given audience, asking insights on tones from other email agents which oversee other clusters. 2. An agent calls the external agent to deploy the campaign. |
 
 <hr />
 
-## Usage
+## Quick Start
+
+**Install `versionhq` package:**
 
-1. Install `versionhq` package:
    ```
-   uv pip install versionhq
+   pip install versionhq
    ```
 
-2. You can use the `versionhq` module in your Python app.
+(Python >= 3.13)
+
 
+### Case 1. Single agent network:
 
-### Case 1. Build an AI agent on LLM of your choice and execute a task:
+#### Return a structured output with a summary in string.
 
    ```
+   from pydantic import BaseModel
    from versionhq.agent.model import Agent
-   from versionhq.task.model import Task, ResponseField
+   from versionhq.task.model import Task
 
-   def my_callback_func():
-      """callback func"""
+   class CustomOutput(BaseModel):
+      test1: str
+      test2: list[str]
+
+   def dummy_func(message: str, test1: str, test2: list[str]) -> str:
+      return f"{message}: {test1}, {", ".join(test2)}"
 
-   agent = Agent(
-      role="demo",
-      goal="amazing project goal",
-      skillsets=["skill_1", "skill_2", ],
-      tools=["amazing RAG tool",]
-      llm="llm-of-your-choice"
-   )
+
+   agent = Agent(role="demo", goal="amazing project goal")
 
    task = Task(
       description="Amazing task",
-      response_fields=[
-         ResponseField(title="test1", data_type=str, required=True),
-         ResponseField(title="test2", data_type=list, items=str, required=True),
-      ],
-      callbacks=[my_callback_func]
+      pydantic_custom_output=CustomOutput,
+      callback=dummy_func,
+      callback_kwargs=dict(message="Hi! Here is the result: ")
    )
+
    res = task.execute_sync(agent=agent, context="amazing context to consider.")
-   return res.to_dict()
+   print(res)
    ```
 
-This will return a dictionary with keys defined in the `ResponseField`.
+This will return `TaskOutput` that stores a response in string, JSON dict, and Pydantic model: `CustomOutput` formats with a callback result.
 
    ```
-   { test1: "answer1", "test2": ["answer2-1", "answer2-2", "answer2-3",] }
+   res == TaskOutput(
+      raw="{\\"test1\\": \\"random str\\", \\"test2\\": [\\"item1\\", \\"item2\\"]}",
+      json_dict={"test1": "random str", "test2": ["item1", "item2"]},
+      pydantic=CustomOutput(test1="random str", test2=["item 1", "item 2"]),
+      callback_output="Hi! Here is the result: random str, item 1, item 2",
+   )
    ```
 
 ### Case 2. Form a team to handle multiple tasks:

{versionhq-1.1.10.5 → versionhq-1.1.10.7}/pyproject.toml

@@ -15,7 +15,7 @@ exclude = ["test*", "__pycache__"]
 
 [project]
 name = "versionhq"
-version = "1.1.10.5"
+version = "1.1.10.7"
 authors = [{ name = "Kuriko Iwai", email = "kuriko@versi0n.io" }]
 description = "LLM orchestration frameworks for model-agnostic AI agents that handle complex outbound workflows"
 readme = "README.md"

{versionhq-1.1.10.5 → versionhq-1.1.10.7}/src/versionhq/__init__.py

@@ -18,7 +18,7 @@ from versionhq.tool.model import Tool
 from versionhq.tool.composio_tool import ComposioHandler
 
 
-__version__ = "1.1.10.5"
+__version__ = "1.1.10.7"
 __all__ = [
     "Agent",
     "Customer",

{versionhq-1.1.10.5 → versionhq-1.1.10.7}/src/versionhq.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.2
 Name: versionhq
-Version: 1.1.10.5
+Version: 1.1.10.7
 Summary: LLM orchestration frameworks for model-agnostic AI agents that handle complex outbound workflows
 Author-email: Kuriko Iwai <kuriko@versi0n.io>
 License: MIT License
@@ -66,32 +66,17 @@ Requires-Dist: composio-langchain>=0.6.12
 ![pyenv ver](https://img.shields.io/badge/pyenv-2.5.0-orange)
 
 
-An LLM orchestration frameworks for multi-agent systems with RAG to autopilot outbound workflows.
-
-Agents are model agnostic.
-
-Messaging workflows are created at individual level, and will be deployed on third-party services via `Composio`.
-
+LLM orchestration frameworks to deploy multi-agent systems focusing on complex outbound tasks.
 
 **Visit:**
 
 - [PyPI](https://pypi.org/project/versionhq/)
-- [Github (LLM orchestration)](https://github.com/versionHQ/multi-agent-system)
-- [Doc](https://chief-oxygen-8a2.notion.site/Documentation-17e923685cf98001a5fad5c4b2acd79b?pvs=4)
-<!-- - [Github (Test client app)](https://github.com/versionHQ/test-client-app) -->
-- [Use case](https://versi0n.io/): Client web app for the outbound messaging automation
+- [Github (LLM orchestration framework)](https://github.com/versionHQ/multi-agent-system)
+- [Github (Test client app)](https://github.com/versionHQ/test-client-app)
+- [Use case](https://versi0n.io/playground) / [Quick demo](https://res.cloudinary.com/dfeirxlea/video/upload/v1737732977/pj_m_home/pnsyh5mfvmilwgt0eusa.mov)
+- [Documentation](https://chief-oxygen-8a2.notion.site/Documentation-17e923685cf98001a5fad5c4b2acd79b?pvs=4) *Some components are under review.
 
 
-<hr />
-
-## Mindmap
-
-LLM-powered `agent`s and `team`s use `tool`s and their own knowledge to complete the `task` given by the client or the system.
-
-<p align="center">
-   <img src="https://res.cloudinary.com/dfeirxlea/image/upload/v1733556715/pj_m_home/urwte15at3h0dr8mdlyo.png" alt="mindmap" width="400">
-</p>
-
 <hr />
 
 ## Table of Content
@@ -99,8 +84,10 @@ LLM-powered `agent`s and `team`s use `tool`s and their own knowledge to complete
 <!-- DON'T EDIT THIS SECTION, INSTEAD RE-RUN doctoc TO UPDATE -->
 
 - [Key Features](#key-features)
-- [Usage](#usage)
-  - [Case 1. Build an AI agent on LLM of your choice and execute a task:](#case-1-build-an-ai-agent-on-llm-of-your-choice-and-execute-a-task)
+  - [Agent formation](#agent-formation)
+- [Quick Start](#quick-start)
+  - [Case 1. Single agent network:](#case-1-single-agent-network)
+    - [Return a structured output with a summary in string.](#return-a-structured-output-with-a-summary-in-string)
   - [Case 2. Form a team to handle multiple tasks:](#case-2-form-a-team-to-handle-multiple-tasks)
 - [Technologies Used](#technologies-used)
 - [Project Structure](#project-structure)
@@ -119,68 +106,75 @@ LLM-powered `agent`s and `team`s use `tool`s and their own knowledge to complete
 
 ## Key Features
 
-A mulit-agent systems with Rag that tailors messaging workflow, predicts its performance, and deploys it on third-party tools.
+Generate mulit-agent systems depending on the complexity of the task, and execute the task with agents of choice.
+
+Model-agnostic agents can handle RAG tools, tools, callbacks, and knowledge sharing among other agents.
 
-The `agent` is model agnostic. The default model is set Chat GTP 4o. We ask the client their preference and switch it accordingly using llm variable stored in the `BaseAgent` class.
 
-Multiple `agents` can form a `team` to complete complex tasks together.
+###  Agent formation
+Depending on the task complexity, agents can make a different formation.
 
-**1. Analysis**
-- Professional `agents` handle the analysis `tasks` on each client, customer, and product.
+You can specify which formation you want them to generate, or let the agent decide if you don’t have a clear plan.
 
-**2. Messaging Workflow Creation**
-- Several `teams` receive the analysis and design initial messaging workflow with several layers.
-- Ask the client for their inputs
-- Deploy the workflow on the third party tools using `composio`.
 
-**3. Autopiloting**
-- Responsible `agents` or `teams` autopilot executing and refining the messaging workflow.
+|  | **Solo Agent** | **Supervising** | **Network** | **Random** |
+| :--- | :--- | :--- | :--- | :--- |
+| **Formation** | <img src="https://res.cloudinary.com/dfeirxlea/image/upload/v1737893140/pj_m_agents/tglrxoiuv7kk7nzvpe1z.jpg" alt="solo" width="200"> | <img src="https://res.cloudinary.com/dfeirxlea/image/upload/v1737893141/pj_m_agents/hxngdvnn5b5qdxo0ayl5.jpg" alt="solo" width="200"> | <img src="https://res.cloudinary.com/dfeirxlea/image/upload/v1737893142/pj_m_agents/kyc6neg8m6keyizszcpi.jpg" alt="solo" width="200"> | <img src="https://res.cloudinary.com/dfeirxlea/image/upload/v1737893139/pj_m_agents/hzpchlcpnpwxwaapu1hr.jpg" alt="solo" width="200"> |
+| **Usage** | <ul><li>A single agent with tools, knowledge, and memory.</li><li>When self-learning mode is on - it will turn into **Random** formation.</li></ul> | <ul><li>Leader agent gives directions, while sharing its knowledge and memory.</li><li>Subordinates can be solo agents or networks.</li></ul> | <ul><li>Share tasks, knowledge, and memory among network members.</li></ul> | <ul><li>A single agent handles tasks, asking help from other agents without sharing its memory or knowledge.</li></ul> |
+| **Use case** | An email agent drafts promo message for the given audience. | The leader agent strategizes an outbound campaign plan and assigns components such as media mix or message creation to subordinate agents. | An email agent and social media agent share the product knowledge and deploy multi-channel outbound campaign. | 1. An email agent drafts promo message for the given audience, asking insights on tones from other email agents which oversee other clusters. 2. An agent calls the external agent to deploy the campaign. |
 
 <hr />
 
-## Usage
+## Quick Start
+
+**Install `versionhq` package:**
 
-1. Install `versionhq` package:
    ```
-   uv pip install versionhq
+   pip install versionhq
    ```
 
-2. You can use the `versionhq` module in your Python app.
+(Python >= 3.13)
+
 
+### Case 1. Single agent network:
 
-### Case 1. Build an AI agent on LLM of your choice and execute a task:
+#### Return a structured output with a summary in string.
 
    ```
+   from pydantic import BaseModel
    from versionhq.agent.model import Agent
-   from versionhq.task.model import Task, ResponseField
+   from versionhq.task.model import Task
 
-   def my_callback_func():
-      """callback func"""
+   class CustomOutput(BaseModel):
+      test1: str
+      test2: list[str]
+
+   def dummy_func(message: str, test1: str, test2: list[str]) -> str:
+      return f"{message}: {test1}, {", ".join(test2)}"
 
-   agent = Agent(
-      role="demo",
-      goal="amazing project goal",
-      skillsets=["skill_1", "skill_2", ],
-      tools=["amazing RAG tool",]
-      llm="llm-of-your-choice"
-   )
+
+   agent = Agent(role="demo", goal="amazing project goal")
 
    task = Task(
       description="Amazing task",
-      response_fields=[
-         ResponseField(title="test1", data_type=str, required=True),
-         ResponseField(title="test2", data_type=list, items=str, required=True),
-      ],
-      callbacks=[my_callback_func]
+      pydantic_custom_output=CustomOutput,
+      callback=dummy_func,
+      callback_kwargs=dict(message="Hi! Here is the result: ")
    )
+
    res = task.execute_sync(agent=agent, context="amazing context to consider.")
-   return res.to_dict()
+   print(res)
    ```
 
-This will return a dictionary with keys defined in the `ResponseField`.
+This will return `TaskOutput` that stores a response in string, JSON dict, and Pydantic model: `CustomOutput` formats with a callback result.
 
    ```
-   { test1: "answer1", "test2": ["answer2-1", "answer2-2", "answer2-3",] }
+   res == TaskOutput(
+      raw="{\\"test1\\": \\"random str\\", \\"test2\\": [\\"item1\\", \\"item2\\"]}",
+      json_dict={"test1": "random str", "test2": ["item1", "item2"]},
+      pydantic=CustomOutput(test1="random str", test2=["item 1", "item 2"]),
+      callback_output="Hi! Here is the result: random str, item 1, item 2",
+   )
    ```
 
 ### Case 2. Form a team to handle multiple tasks:

{versionhq-1.1.10.5 → versionhq-1.1.10.7}/tests/task/task_test.py

@@ -126,8 +126,6 @@ def test_callback():
     def callback_func(kwargs: Dict[str, Any]):
         task_id = kwargs.get("task_id", None)
         added_condition = kwargs.get("added_condition", None)
-        assert task_id is not None
-        assert added_condition is not None
         return f"Result: {task_id}, condition added: {added_condition}"
 
     task = Task(
@@ -140,10 +138,9 @@ def test_callback():
     )
     res = task.execute_sync(agent=base_agent)
 
-    assert res is not None
-    assert isinstance(res, TaskOutput)
+    assert res and isinstance(res, TaskOutput)
     assert res.task_id is task.id
-    assert res.raw is not None
+    assert "demo for pytest" in res.callback_output
 
 
 def test_delegate():
@@ -303,6 +300,8 @@ def test_callback():
 
 
 def test_task_with_agent_callback():
+    import litellm
+
     def dummy_func(*args, **kwargs) -> str:
         return "Demo func"
 
@@ -311,6 +310,7 @@ def test_task_with_agent_callback():
     res = task.execute_sync(agent=agent)
 
     assert res.raw and res.task_id == task.id
+    assert litellm.callbacks == [dummy_func]
 
 # task - maxit, loop, rpm