versionhq 1.1.7.0__py3-none-any.whl → 1.1.7.2__py3-none-any.whl

versionhq/__init__.py

@@ -17,7 +17,7 @@ from versionhq.team.model import Team, TeamOutput
 from versionhq.tool.model import Tool
 
 
-__version__ = "1.1.7.0"
+__version__ = "1.1.7.2"
 __all__ = [
     "Agent",
     "Customer",

versionhq/agent/TEMPLATES/Backstory.py

@@ -1,7 +1,3 @@
 BACKSTORY="""You are a {role} with deep understanding of {knowledge} and highly skilled in {skillsets}.
 You have access to call the RAG tools that can {rag_tool_overview}. By leveraging these tools, your knowledge, and skillsets, you can identify competitive strategies that have been proven effective to achieve the goal: {goal}. Take these into consideration, create innovative solutions.
 """
-
-
-# developing strategies to improve cohort retention and customer lifetime value
-# ingest past successful campaigns

versionhq/agent/model.py

@@ -367,8 +367,8 @@ class Agent(ABC, BaseModel):
         """
 
         task_prompt = task.prompt()
-        # if context:
-        #     task_prompt = self.i18n.slice("task_with_context").format(task=task_prompt, context=context)
+        if context is not task.prompt_context:  # as `task.prompt()` includes adding `task.prompt_context` to the prompt.
+            task_prompt += context
 
         tool_results = []
         if task.tools_called:

versionhq/task/model.py

@@ -18,21 +18,44 @@ class ResponseField(BaseModel):
     """
     Field class to use in the response schema for the JSON response.
     """
+
     title: str = Field(default=None)
     type: Type = Field(default=str)
     required: bool = Field(default=True)
 
+
     def _annotate(self, value: Any) -> Annotated:
         """
         Address `create_model`
         """
         return Annotated[self.type, value] if isinstance(value, self.type) else Annotated[str, str(value)]
 
+
+    def _convert(self, value: Any) -> Any:
+        try:
+            if self.type is Any:
+                pass
+            elif self.type is int:
+                return int(value)
+            elif self.type is float:
+                return float(value)
+            elif self.type is list or self.type is dict:
+                return json.loads(value)
+            else:
+                return value
+        except:
+            return value
+
+
     def create_pydantic_model(self, result: Dict, base_model: Union[BaseModel | Any]) -> Any:
         for k, v in result.items():
-            if k is not self.title or type(v) is not self.type:
+            if k is not self.title:
                 pass
-            setattr(base_model, k, v)
+            elif type(v) is not self.type:
+                v = self._convert(v)
+                setattr(base_model, k, v)
+            else:
+                setattr(base_model, k, v)
         return base_model
 
 
@@ -43,12 +66,14 @@ class AgentOutput(BaseModel):
     """
     customer_id: str = Field(default=None, max_length=126, description="customer uuid")
     customer_analysis: str = Field(default=None, max_length=256, description="analysis of the customer")
-    business_overview: str = Field(default=None,max_length=256,description="analysis of the client's business")
-    cohort_timeframe: int = Field(default=None,max_length=256,description="Suitable cohort timeframe in days")
+    product_overview: str = Field(default=None, max_length=256, description="analysis of the client's business")
+    usp: str = Field()
+    cohort_timeframe: int = Field(default=None, max_length=256, description="suitable cohort timeframe in days")
     kpi_metrics: List[str] = Field(default=list, description="Ideal KPIs to be tracked")
     assumptions: List[Dict[str, Any]] = Field(default=list, description="assumptions to test")
 
 
+
 class TaskOutput(BaseModel):
     """
     Store the final output of the task in TaskOutput class.
@@ -57,8 +82,8 @@ class TaskOutput(BaseModel):
 
     task_id: UUID4 = Field(default_factory=uuid.uuid4, frozen=True, description="store Task ID")
     raw: str = Field(default="", description="Raw output of the task")
-    pydantic: Optional[Any] = Field(default=None, description="`raw` converted to the abs. pydantic model")
     json_dict: Union[Dict[str, Any]] = Field(default=None, description="`raw` converted to dictionary")
+    pydantic: Optional[Any] = Field(default=None, description="`raw` converted to the abs. pydantic model")
 
     def __str__(self) -> str:
         return str(self.pydantic) if self.pydantic else str(self.json_dict) if self.json_dict else self.raw
@@ -75,16 +100,31 @@ class TaskOutput(BaseModel):
             )
         return json.dumps(self.json_dict)
 
+
     def to_dict(self) -> Dict[str, Any]:
-        """Convert json_output and pydantic_output to a dictionary."""
+        """
+        Convert pydantic / raw output into dict and return the dict.
+        When we only have `raw` output, return `{ output: raw }` to avoid an error
+        """
+
         output_dict = {}
         if self.json_dict:
             output_dict.update(self.json_dict)
         elif self.pydantic:
             output_dict.update(self.pydantic.model_dump())
+        else:
+            output_dict.upate({ "output": self.raw })
         return output_dict
 
 
+    def context_prompting(self) -> str:
+        """
+        When the task is called as context, return its output in concise string to add it to the prompt
+        """
+        return json.dumps(self.json_dict) if self.json_dict else self.raw[0: 127]
+
+
+
 class Task(BaseModel):
     """
     Task to be executed by the agent or the team.
@@ -102,7 +142,10 @@ class Task(BaseModel):
     # output
     expected_output_json: bool = Field(default=True)
     expected_output_pydantic: bool = Field(default=False)
-    output_field_list: Optional[List[ResponseField]] = Field(default=[ResponseField(title="output")])
+    output_field_list: List[ResponseField] = Field(
+        default=[ResponseField(title="output", type=str, required=False)],
+        description="provide output key and data type. this will be cascaded to the agent via task.prompt()"
+    )
     output: Optional[TaskOutput] = Field(default=None, description="store the final task output in TaskOutput class")
 
     # task setup
@@ -110,7 +153,7 @@ class Task(BaseModel):
     tools_called: Optional[List[ToolCalled]] = Field(default_factory=list, description="tools that the agent can use for this task")
     take_tool_res_as_final: bool = Field(default=False,description="when set True, tools res will be stored in the `TaskOutput`")
 
-    prompt_context: Optional[str] = None
+    prompt_context: Optional[str] = Field(default=None)
     async_execution: bool = Field(default=False,description="whether the task should be executed asynchronously or not")
     config: Optional[Dict[str, Any]] = Field(default=None, description="configuration for the agent")
     callback: Optional[Any] = Field(default=None, description="callback to be executed after the task is completed.")
@@ -123,18 +166,18 @@ class Task(BaseModel):
 
 
     @property
-    def output_prompt(self):
+    def output_prompt(self) -> str:
         """
         Draft prompts on the output format by converting `output_field_list` to dictionary.
         """
 
-        output_prompt, output_dict = "", dict()
+        output_prompt, output_formats_to_follow = "", dict()
         for item in self.output_field_list:
-            output_dict[item.title] = f"<Return your answer in {item.type.__name__}>"
+            output_formats_to_follow[item.title] = f"<Return your answer in {item.type.__name__}>"
 
         output_prompt = f"""
-        Your outputs STRICTLY follow the following format and should NOT contain any other irrevant elements that not specified in the following format:
-        {output_dict}
+Your outputs MUST adhere to the following format and should NOT include any irrelevant elements:
+{output_formats_to_follow}
         """
         return output_prompt
 
@@ -228,16 +271,28 @@ class Task(BaseModel):
         return self
 
 
-    def prompt(self, customer=str | None, product_overview=str | None) -> str:
+    def prompt(self, customer: str = None, product_overview: str = None) -> str:
         """
-        Format the task prompt.
+        Format the task prompt and cascade it to the agent.
+        When the task has context, add context prompting of all the tasks in the context.
+        When we have cusotmer/product info, add them to the prompt.
         """
-        task_slices = [
-            self.description,
-            f"Customer overview: {customer}",
-            f"Product overview: {product_overview}",
-            f"{self.output_prompt}",
-        ]
+
+        task_slices = [self.description, f"{self.output_prompt}", f"Take the following context into consideration: "]
+
+        if self.context:
+            context_outputs = "\n".join([task.output.context_prompting() if hasattr(task, "output") else "" for task in self.context])
+            task_slices.insert(len(task_slices),  context_outputs)
+
+        if customer:
+            task_slices.insert(len(task_slices),  f"Customer overview: {customer}")
+
+        if product_overview:
+            task_slices.insert(len(task_slices), f"Product overview: {product_overview}")
+
+        if self.prompt_context:
+            task_slices.insert(len(task_slices), self.prompt_context)
+
         return "\n".join(task_slices)
 
 
@@ -273,10 +328,10 @@ class Task(BaseModel):
 
     def create_pydantic_output(self, output_json_dict: Dict[str, Any], raw_result: Any = None) -> Optional[Any]:
         """
-        Create pydantic output from the raw result.
+        Create pydantic output from the `raw` result.
         """
 
-        output_pydantic = None #! REFINEME
+        output_pydantic = None
         if isinstance(raw_result, BaseModel):
             output_pydantic = raw_result
 
@@ -285,8 +340,14 @@ class Task(BaseModel):
 
         else:
             output_pydantic = create_model("PydanticTaskOutput", __base__=BaseModel)
-            for item in self.output_field_list:
-                item.create_pydantic_model(result=output_json_dict, base_model=output_pydantic)
+            try:
+                for item in self.output_field_list:
+                    value = output_json_dict[item.title] if hasattr(output_json_dict, item.title) else None
+                    if value and type(value) is not item.type:
+                        value = item._convert(value)
+                    setattr(output_pydantic, item.title, value)
+            except:
+                setattr(output_pydantic, "output", output_json_dict)
 
         return output_pydantic
 
@@ -307,11 +368,19 @@ class Task(BaseModel):
             self.description = self._original_description.format(**inputs)
             # self.expected_output = self._original_expected_output.format(**inputs)
 
+
     # task execution
     def execute_sync(self, agent, context: Optional[str] = None) -> TaskOutput:
         """
         Execute the task synchronously.
+        When the task has context, make sure we have executed all the tasks in the context first.
         """
+
+        if self.context:
+            for task in self.context:
+                if task.output is None:
+                    task._execute_core(agent, context)
+
         return self._execute_core(agent, context)
 
 
@@ -341,12 +410,12 @@ class Task(BaseModel):
         """
 
         self.prompt_context = context
-        raw_result = agent.execute_task(task=self, context=context)
-        output_json_dict = self.create_json_output(raw_result=raw_result)
+        output_raw = agent.execute_task(task=self, context=context)
+        output_json_dict = self.create_json_output(raw_result=output_raw)
         output_pydantic = self.create_pydantic_output(output_json_dict=output_json_dict)
         task_output = TaskOutput(
             task_id=self.id,
-            raw=raw_result,
+            raw=output_raw,
             pydantic=output_pydantic,
             json_dict=output_json_dict
         )

versionhq/team/model.py

@@ -100,26 +100,27 @@ class TeamOutput(BaseModel):
 
     def to_dict(self) -> Dict[str, Any]:
         """
-        Convert json_output and pydantic_output to a dictionary.
+        Convert pydantic / raw output into dict and return the dict.
+        When we only have `raw` output, return `{ output: raw }` to avoid an error
         """
+
         output_dict = {}
         if self.json_dict:
             output_dict.update(self.json_dict)
         elif self.pydantic:
             output_dict.update(self.pydantic.model_dump())
         else:
-            output_dict.update({"output", self.raw})
+            output_dict.upate({ "output": self.raw })
         return output_dict
 
+
     def return_all_task_outputs(self) -> List[Dict[str, Any]]:
         res = [output.json_dict for output in self.task_output_list]
         return res
 
 
 class TeamMember(ABC, BaseModel):
-    agent: Agent | None = Field(
-        default=None, description="store the agent to be a member"
-    )
+    agent: Agent | None = Field(default=None, description="store the agent to be a member")
     is_manager: bool = Field(default=False)
     task: Task | None = Field(default=None)
 
@@ -164,9 +165,11 @@ class Team(BaseModel):
     execution_logs: List[Dict[str, Any]] = Field(default=[], description="list of execution logs for tasks")
     usage_metrics: Optional[UsageMetrics] = Field(default=None, description="usage metrics for all the llm executions")
 
+
     def __name__(self) -> str:
         return self.name if self.name is not None else self.id.__str__
 
+
     @property
     def key(self) -> str:
         source = [str(member.agent.id.__str__) for member in self.members] + [str(task.id.__str__) for task in self.tasks]
@@ -175,9 +178,7 @@ class Team(BaseModel):
 
     @property
     def manager_agent(self) -> Agent:
-        manager_agent = [
-            member.agent for member in self.members if member.is_manager == True
-        ]
+        manager_agent = [member.agent for member in self.members if member.is_manager == True]
         return manager_agent[0] if len(manager_agent) > 0 else None
 
 
@@ -251,7 +252,6 @@ class Team(BaseModel):
         """
         Every team member should have a task to handle.
         """
-
         if self.process == TaskHandlingProcess.sequential:
             for member in self.members:
                 if member.task is None:

{versionhq-1.1.7.0.dist-info → versionhq-1.1.7.2.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: versionhq
-Version: 1.1.7.0
+Version: 1.1.7.2
 Summary: LLM orchestration frameworks for model-agnostic AI agents that handle complex outbound workflows
 Author-email: Kuriko Iwai <kuriko@versi0n.io>
 License: MIT License
@@ -52,8 +52,7 @@ Requires-Dist: wheel>=0.45.1
 
 # Overview
 
-![MIT license](https://img.shields.io/badge/License-MIT-green) ![PyPi](https://img.shields.io/badge/pypi-v1.1.6.3-blue)
-![python ver](https://img.shields.io/badge/Python-3.12/3.13-purple) ![pyenv ver](https://img.shields.io/badge/pyenv-2.4.23-orange)
+![MIT license](https://img.shields.io/badge/License-MIT-green) [![Publisher](https://github.com/versionHQ/multi-agent-system/actions/workflows/publish.yml/badge.svg)](https://github.com/versionHQ/multi-agent-system/actions/workflows/publish.yml) ![PyPi](https://img.shields.io/badge/pypi-v1.1.7.0-blue) ![python ver](https://img.shields.io/badge/Python-3.12/3.13-purple) ![pyenv ver](https://img.shields.io/badge/pyenv-2.4.23-orange)
 
 
 An LLM orchestration frameworks for multi-agent systems with RAG to autopilot outbound workflows.
@@ -70,6 +69,7 @@ Messaging workflows are created at individual level, and will be deployed on thi
 - [Orchestration frameworks](https://github.com/versionHQ/multi-agent-system)
 - [Test client app](https://github.com/versionHQ/test-client-app)
 
+<hr />
 
 ## Mindmap
 
@@ -86,10 +86,10 @@ 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)
 - [Technologies Used](#technologies-used)
 - [Project Structure](#project-structure)
 - [Setup](#setup)
-- [Usage](#usage)
 - [Contributing](#contributing)
   - [Customizing AI Agents](#customizing-ai-agents)
   - [Modifying RAG Functionality](#modifying-rag-functionality)
@@ -121,6 +121,52 @@ Multiple `agents` can form a `team` to complete complex tasks together.
 **3. Autopiloting**
 - Responsible `agents` or `teams` autopilot executing and refining the messaging workflow.
 
+<hr />
+
+## Usage
+
+1. Install `versionhq` package:
+   ```
+   uv pip install versionhq
+   ```
+
+2. You can use the `versionhq` module in your Python app.
+
+   - **i.e.,** Make LLM-based agent execute the task and return JSON dict.
+
+   ```
+   from versionhq.agent.model import Agent
+   from versionhq.task.model import Task, ResponseField
+
+   agent = Agent(
+      role="demo",
+      goal="amazing project goal",
+      skillsets=["skill_1", "skill_2", ],
+      llm="llm-of-choice"
+   )
+
+   task = Task(
+      description="Amazing task",
+      expected_output_json=True,
+      expected_output_pydantic=False,
+      output_field_list=[
+         ResponseField(title="test1", type=str, required=True),
+         ResponseField(title="test2", type=list, required=True),
+      ],
+      context=["amazing context",],
+      tools=["amazing tool"],
+      callback=None,
+   )
+
+   res = task.execute_sync(agent=agent)
+
+   return res.to_dict()
+
+   ```
+
+For more details:
+
+[PyPi package](https://pypi.org/project/versionhq/)
 
 <hr />
 
@@ -145,6 +191,7 @@ Multiple `agents` can form a `team` to complete complex tasks together.
    - [pre-commit](https://pre-commit.com/): Manage and maintain pre-commit hooks
    - [setuptools](https://pypi.org/project/setuptools/): Build python modules
 
+<hr />
 
 ## Project Structure
 
@@ -210,25 +257,6 @@ src/
 
 <hr />
 
-## Usage
-
-1. Install `versionhq` package:
-```
-uv pip install versionhq
-```
-
-2. You can use the `versionhq` module in your Python app.
-```
-from versionhq.agent.model import Agent
-agent = Agent(llm="your-llm", ...)
-```
-
-For more details:
-
-[PyPi package](https://pypi.org/project/versionhq/)
-
-<hr />
-
 ## Contributing
 
 1. Fork the repository
@@ -359,17 +387,3 @@ Common issues and solutions:
 > **Simple tasks**: You have a straightforward, one-off task that doesn't require significant complexity or iteration.
 
 > **Human input**: You need to provide initial input or guidance to the agent, or you expect to review and refine the output.
-
-
-
-<--- Remaining tasks --->
-
-- llm handling - agent
-- more llms integration
-- simpler prompting
-- broader knowledge
-
-- utils - log
-- utils - time
-
-- end to end client app test

{versionhq-1.1.7.0.dist-info → versionhq-1.1.7.2.dist-info}/RECORD

@@ -1,4 +1,4 @@
-versionhq/__init__.py,sha256=9ZXfGKk15KYd9cggWAR006GP5BGegF63yq4jLwQwFyA,773
+versionhq/__init__.py,sha256=ZRUf-j8kvlaLsgzQHU-EVzvct-p6Rrrm8X7wLmQcgrY,773
 versionhq/_utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 versionhq/_utils/cache_handler.py,sha256=zDQKzIn7vp-M2-uepHFxgJstjfftZS5mzXKL_-4uVvI,370
 versionhq/_utils/i18n.py,sha256=TwA_PnYfDLA6VqlUDPuybdV9lgi3Frh_ASsb_X8jJo8,1483
@@ -7,9 +7,9 @@ versionhq/_utils/process_config.py,sha256=ogrhovLbwe0ocQlcohRgBBRtww7C3pk9hikjvg
 versionhq/_utils/rpm_controller.py,sha256=T7waIGeblu5K58erY4lqVLcPsWM7W9UFdU3DG9Dsk0w,2214
 versionhq/_utils/usage_metrics.py,sha256=c33a_28y8ECUgflsKN3mkNm0fNkWgZmXwybMwIqoKXA,1098
 versionhq/agent/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-versionhq/agent/model.py,sha256=Xp_S57uMa_R2PNN7XOvJQWUUDiPIuSC3leN8tVbjWSQ,18331
+versionhq/agent/model.py,sha256=nJT3EOkFAEV70fcEeNS8ctkzztONK-HhwMkY39qtEzo,18359
 versionhq/agent/parser.py,sha256=GhoNQo4WloVM3vGnAmt9lnEOTARX7nWMhJE55rF_5Rs,5500
-versionhq/agent/TEMPLATES/Backstory.py,sha256=Sh6KSdKVkWvwfdHWxbSjLIOEbj83LBWa3GVDbET3-FE,512
+versionhq/agent/TEMPLATES/Backstory.py,sha256=cdngBx1GEv7nroR46FEhnysnBJ9mEVL763_9np6Skkc,395
 versionhq/agent/TEMPLATES/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 versionhq/cli/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 versionhq/clients/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -24,16 +24,16 @@ versionhq/llm/llm_vars.py,sha256=YZoXqFBW7XpclUZ14_AAz7WOjoyCXnGcI959GSpX2q0,534
 versionhq/llm/model.py,sha256=PdwisrlrsDqd6gXwXCyGbGTRTeGZ8SXpt_gfua8qunk,8266
 versionhq/task/__init__.py,sha256=g4mCATnn1mUXxsfQ5p6IpPawr8O421wVIT8kMKEcxQw,180
 versionhq/task/formatter.py,sha256=N8Kmk9vtrMtBdgJ8J7RmlKNMdZWSmV8O1bDexmCWgU0,643
-versionhq/task/model.py,sha256=avhGv1FswgHPhnZ_JisqhXT_dDNosarDBKYxVmA28PI,14633
+versionhq/task/model.py,sha256=maWDeVm1CfqGw-re42a02-0SXtCmkjXsz_ociB1epfQ,17000
 versionhq/team/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-versionhq/team/model.py,sha256=h5WUmF7yXYLaG9gYZJEMvjZsP5qkHpk5qUvfbPsYonY,20085
+versionhq/team/model.py,sha256=pMCt0ZCVgE_pRH2qXlWj-PEZrI77JoRVVzUOZWY1Rws,20140
 versionhq/team/team_planner.py,sha256=B1UOn_DYVVterUn2CAd80jfO4sViJCCXPJA3abSSugg,2143
 versionhq/tool/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 versionhq/tool/decorator.py,sha256=Y-j4jkoujD5LUvpe8uf3p5Zagk2XVaRKC9rkIE-2geo,1189
 versionhq/tool/model.py,sha256=JZOEcZRIEfcrjL8DgrFYDt4YNgMF8rXS26RK6D2x9mc,6906
 versionhq/tool/tool_handler.py,sha256=e-2VfG9zFpfPG_oMoPXye93GDovs7FuUASWQwUTLrJ0,1498
-versionhq-1.1.7.0.dist-info/LICENSE,sha256=7CCXuMrAjPVsUvZrsBq9DsxI2rLDUSYXR_qj4yO_ZII,1077
-versionhq-1.1.7.0.dist-info/METADATA,sha256=11jMeAq7XDlWkqYtfwHCitC7LQb5Z2suhol9ekX6g3U,13552
-versionhq-1.1.7.0.dist-info/WHEEL,sha256=PZUExdf71Ui_so67QXpySuHtCi3-J3wvF4ORK6k_S8U,91
-versionhq-1.1.7.0.dist-info/top_level.txt,sha256=DClQwxDWqIUGeRJkA8vBlgeNsYZs4_nJWMonzFt5Wj0,10
-versionhq-1.1.7.0.dist-info/RECORD,,
+versionhq-1.1.7.2.dist-info/LICENSE,sha256=7CCXuMrAjPVsUvZrsBq9DsxI2rLDUSYXR_qj4yO_ZII,1077
+versionhq-1.1.7.2.dist-info/METADATA,sha256=6xxLMRkGhlbN1siAlvNsRQc1JWGTk0hIT4J5ziNMaZQ,14278
+versionhq-1.1.7.2.dist-info/WHEEL,sha256=PZUExdf71Ui_so67QXpySuHtCi3-J3wvF4ORK6k_S8U,91
+versionhq-1.1.7.2.dist-info/top_level.txt,sha256=DClQwxDWqIUGeRJkA8vBlgeNsYZs4_nJWMonzFt5Wj0,10
+versionhq-1.1.7.2.dist-info/RECORD,,