hammad-python

Harmoniously Accelerated Micro-Modules (for) Application Development

---

hammad-python is an ecosystem of wrappers with the sole purpose of defining opinionated resources that I myself use in my projects and day to day work when building with Python.

The ecosystem is split into 4 main sub-packages:

• hammad-python-core : Contains "stdlib" like resources such as converters, types, caching, etc. All of these resources can easily be implemented from scratch, and are built soley for convenience.
• hammad-python-data : Contains a Collection class / system that implements a simple vector / non vector database system.
• hammad-python-genai : Contains a collection of patterns for using Generative AI models as well as implementing concepts like Agentic Reasoning.
• hammad-python-http : HTTP / API related resources and utilities. Contains resources related to MCP, HTTP, OpenAPI & GraphQL Servers.

---

Installation

You can install any of the sub-packages, or the main package through pip or uv.

pip install hammad-python

---

Quickstart

Easily Use Language Models With Various Tools & Patterns

from ham.genai import Agent, LanguageModel
from ham.http import SearchClient

client = SearchClient()

def search_web(query: str) -> str:
    return client.search(query)

# create your agent
# agents have rich functionality, and use `litellm` & `instructor` under the hood
agent = Agent(
    instructions = "You are a helpful assistant who can search the web",
    tools = [search_web]
)

print(agent.run("What is the weather in Tokyo?"))

>>> AgentResponse:
city='Tokyo' temperature=25.0 description='Clear sky'

>>> Model: openai/gpt-4o-mini
>>> Steps: 2
>>> Output Type: Response
>>> Total Tool Calls: 1

Vectorized & Searchable Collections

from ham.data import Collection

collection = Collection(vector=True)

collection.add("Rocks")
collection.add("Cucumber")
collection.add("Dragon")
collection.add("Apple")

# query the collection
for result in collection.query("Fruit"):
    print(result.item)

>>> Apple
>>> Cucumber
>>> Dragon
>>> Rocks

HTTP & API Resources

from ham.http import function_server

# instantly launch a server by decorating a function
@function_server(path="/some-endpoint", auto_start=True)
def some_endpoint(name : str) -> str:
    return f"Hello, {name}!"

Now in a separate file:

import requests

print(
    requests.post(
        "http://0.0.0.0:8000/some-endpoint",
        json = {"name": "John"}
    ).json()
)

>>> {'result' : 'Hello, John!'}