Prompt Builder
Library for building prompts and agents with LLMs.
Installation
From PyPI:
pip install promptbuilder
From source:
git clone https://github.com/kapulkin/promptbuilder.git
cd promptbuilder
pip install -e .
Features
- Prompt templates with variables and content tags
- Structured output with TypeScript-like schema definition
- LLM client with native structured output support and caching option
- Integration with multiple LLM providers through aisuite
- Agents with routing based on tools
- Tools as agent for flexibility and scalability
Quick Start
Basic Prompt Usage
from promptbuilder.llm_client import LLMClient
from promptbuilder.prompt_builder import PromptBuilder
# Build prompt template
prompt_template = PromptBuilder() \
.text("What is the capital of ").variable("country").text("?") \
.build()
# Use with LLM
llm_client = LLMClient(model="your-model", api_key="your-api-key")
response = llm_client.from_text(
prompt_template.render(country="France")
)
print(response)
Using Agents
from typing import List
from pydantic import BaseModel, Field
from promptbuilder.agent.agent import AgentRouter
from promptbuilder.agent.context import Context, InMemoryDialogHistory
from promptbuilder.agent.message import Message
from promptbuilder.llm_client import LLMClient
# Define tool arguments
class AddTodoArgs(BaseModel):
item: TodoItem = Field(..., description="Todo item to add")
# Create custom context
class TodoItem(BaseModel):
description: str = Field(..., description="Description of the todo item")
class TodoListContext(Context[InMemoryDialogHistory]):
todos: List[TodoItem] = []
# Create agent with tools
class TodoListAgent(AgentRouter[InMemoryDialogHistory, TodoListContext]):
def __init__(self, llm_client: LLMClient, context: TodoListContext):
super().__init__(llm_client=llm_client, context=context)
llm_client = LLMClient(model="your-model", api_key="your-api-key")
agent = TodoListAgent(llm_client=llm_client, context=TodoListContext())
@agent.tool(description="Add a new todo item to the list", args_model=AddTodoArgs)
async def add_todo(message: Message, args: AddTodoArgs, context: TodoListContext) -> str:
context.todos.append(args.item)
return f"Added todo item: {args.item.description}"
# Use the agent
async def main():
response = await agent(Message(role="user", content="Add a todo: Buy groceries"))
print(response)
See the examples directory for more detailed examples, including a complete todo list manager.
License
MIT License - see LICENSE file for details.
Contributing
Contributions are welcome! Please feel free to submit a pull request.
1 maintainer
Metadata
1 maintainer
Release files for promptbuilder 0.5.5
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| promptbuilder-0.5.5.tar.gz | 54.8 kB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| promptbuilder-0.5.5-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 110.0 kB
Release files / promptbuilder-0.5.5.tar.gz
| Download URL | promptbuilder-0.5.5.tar.gz |
|---|---|
| Size | 54.8 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
cbb773e181adb43d958034dfea0c9f7786527cd7c6c317b89ae28ba8f2bbb277
|
|
BLAKE2b-256 checksum How to use checksums |
20a1d152a2a00097ecb389df0e324ad74bad224f46b9304b686b0edb135971a8
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.1.0 CPython/3.13.5
|
Release files / promptbuilder-0.5.5-py3-none-any.whl
| Download URL | promptbuilder-0.5.5-py3-none-any.whl |
|---|---|
| Size | 55.1 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
705a88d24428f13cc25c0797ab26a6b16bef1817fbdab10c6f255ba3363efa03
|
|
BLAKE2b-256 checksum How to use checksums |
c2860caca929a6f13e829f4f14dd71e044e8390642c50fb0fd510154083a98d5
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.1.0 CPython/3.13.5
|
