infeready 0.0.1

Inferready helps you combine all the components of your prompt into a single LLM-ready payload

infeready

Infeready helps you craft efficient LLM-ready prompts from smaller building blocks. Add history, data sources, ICL examples, and apply filters such as prompt injection scanning, PII redaction or access control over the entire input.

Quick Install

With pip:

pip install infeready

Examples

Example invocation showcasing varying sources and filters (not all of which may be implemented at the time)

from infeready import user_prompt, SystemPrompt
from infeready.sources import PDF, GoogleDrive, LangchainDocuments
from infeready.filters import PromptInject, Anonymize, AccessControl

examples = infeready.load_examples("./icl_examples.json")

system = SystemPrompt("You are LawyerGPT, and must estimate the most likely judicial outcome for a user provided case. Consider all provided documents and respond concisely."}

prompt = user_prompt(
    "determine the most likely legal outcome for the provided case",
    history=[system],
    examples=examples,
    sources=[
        PDF('case_file.pdf'),
        GoogleDrive("credentials.json")
        LangchainDocuments(get_langchain_docs()) # DIY fetch from vectorstore
    ],
    filters=[PromptInject(provider="epoch"), Anonymize(provider="local")],
    max_token_count=10000
)

# We can use the prompt in any LLM or library format
from openai import OpenAI
client = OpenAI()
stream = client.chat.completions.create(
    model="gpt-4",
    messages=prompt.to_openai_messages(),
    stream=True,
)

Example invocation with the openai messages format

from infeready import messages_prompt
from infeready.sources import PDF, GoogleDrive, FromDocuments
from infeready.filters import PromptInject, Anonymize, AccessControl

examples = infeready.load_examples("./icl_examples.json")

messages = [
    {
        "role": "system",
        "content": "You are LawyerGPT, and must estimate the most likely judicial outcome for a user provided case. Consider all provided documents and respond concisely."
    },
    {
        "role": "user",
        "content": "determine the most likely legal outcome for the provided case"
    }
]

prompt = messages_prompt(
    messages,
    examples=examples,
    sources=[
        PDF('case_file.pdf'),
        GoogleDrive("credentials.json")
        FromDocuments(get_langchain_docs()) # DIY fetch from vectorstore
    ],
    filters=[PromptInject(provider="epoch"), Anonymize(provider="local")],
    max_token_count=10000
)


# Use the prompt with langchain
from langchain_openai import ChatOpenAI

model = ChatOpenAI()
chain = prompt.to_langchain_messages() | model
chain.invoke()

Features

• Unit test creation: test prompt creation consistently every time
• Security and privacy scanners
• Compatible with openai and langchain input formats