truthstack 1.0.0

TruthStack supplement safety API SDK. LangChain, LlamaIndex, CrewAI, and OpenAI integrations.

truthstack

Python SDK for TruthStack supplement safety API. Works with LangChain, LlamaIndex, CrewAI, OpenAI function calling, and standalone.

Install

pip install truthstack                    # Core client (zero deps)
pip install truthstack[langchain]         # + LangChain tools
pip install truthstack[llamaindex]        # + LlamaIndex tools
pip install truthstack[crewai]            # + CrewAI tools
pip install truthstack[all]              # Everything

Quick Start

Standalone

from truthstack import TruthStackClient

client = TruthStackClient(api_key="your-key")

# Check interactions (the money endpoint)
result = client.check_interactions(
    supplements=["berberine", "magnesium", "vitamin_d"],
    medications=["semaglutide"]
)
print(result["risk_level"])  # "HIGH"
print(result["cyp_pathway_conflicts"])
print(result["evidence_by_compound"])

# Get evidence balance
evidence = client.get_evidence("berberine")
print(evidence["evidence_balance"])  # {"positive": 8, "null": 2, ...}
print(evidence["citations"])  # [{"pmid": "12345", ...}]

LangChain

from truthstack import get_langchain_tools
from langchain_openai import ChatOpenAI
from langchain.agents import AgentExecutor, create_openai_tools_agent

tools = get_langchain_tools(api_key="your-key")
llm = ChatOpenAI(model="gpt-4")
agent = create_openai_tools_agent(llm, tools, prompt)
executor = AgentExecutor(agent=agent, tools=tools)

result = executor.invoke({"input": "Can I take berberine while on Ozempic?"})

LlamaIndex

from truthstack import get_llamaindex_tools
from llama_index.core.agent import ReActAgent
from llama_index.llms.openai import OpenAI

tools = get_llamaindex_tools(api_key="your-key")
agent = ReActAgent.from_tools(tools, llm=OpenAI(model="gpt-4"))

response = agent.chat("What's the evidence for CoQ10?")

CrewAI

from truthstack import get_crewai_tools
from crewai import Agent, Task, Crew

tools = get_crewai_tools(api_key="your-key")

analyst = Agent(
    role="Supplement Safety Analyst",
    goal="Check supplement interactions and evidence",
    tools=tools,
)

OpenAI Function Calling

from truthstack import get_openai_functions, handle_openai_function_call
from openai import OpenAI

client = OpenAI()
functions = get_openai_functions()

response = client.chat.completions.create(
    model="gpt-4",
    messages=[{"role": "user", "content": "Is magnesium safe with sertraline?"}],
    functions=functions,
    function_call="auto",
)

call = response.choices[0].message.function_call
result = handle_openai_function_call(
    call.name, json.loads(call.arguments), api_key="your-key"
)

System Prompts

from truthstack import SYSTEM_PROMPTS

SYSTEM_PROMPTS["supplement_safety_advisor"]   # General safety
SYSTEM_PROMPTS["glp1_supplement_checker"]     # GLP-1/Ozempic focus
SYSTEM_PROMPTS["evidence_researcher"]         # Research analysis
SYSTEM_PROMPTS["stack_optimizer"]             # Stack optimization
SYSTEM_PROMPTS["clinician_assistant"]         # Clinician-facing

Tools

Tool	Description
truthstack_check_interactions	Check supplement + drug interactions
truthstack_get_evidence	Evidence balance + citations
truthstack_search_supplements	Fuzzy compound search
truthstack_safety_signals	FDA CAERS/FAERS signals
truthstack_drug_profile	Drug CYP450 + botanical interactions

Environment Variables

TRUTHSTACK_API_KEY=your-key
TRUTHSTACK_BASE_URL=https://api.truthstack.co  # optional

API Key

Get a free key (2,000 calls/month) at truthstack.co/developers.

License

MIT