Clearprüf

Know What Your AI Is Doing

Clearprüf is the control layer for agentic AI — monitor behavior, enforce policies, and prevent drift.

Join the waitlist

Your agent has no audit trail.

You can't prove it's safe.

You don't know when its behavior changes.

You don't control what it accesses.

How It Works

Wraps around GPT agents or LangChain flows

Integrate with a single line of code. No architecture changes required.

Logs every action and decision

Complete visibility into agent behavior with detailed audit trails.

Flags risky behavior and violations

Enforces policies like "no file uploads" or "no PII in outputs".

Simple Integration

Add control to your existing agents with minimal code changes.

agent.py

from clearpruf import wrap
from langchain.agents import initialize_agent, load_tools
from langchain.chat_models import ChatOpenAI

llm = ChatOpenAI(model="gpt-4", temperature=0)
tools = load_tools(["serpapi", "requests"])
raw_agent = initialize_agent(tools, llm, agent="zero-shot-react-description")

agent = wrap(raw_agent, config={
    "log": True,
    "enforce": {
        "no_file_uploads": True,
        "no_pii_outputs": True
    },
    "alert": {
        "on_prompt_injection": True,
        "on_unusual_behavior": True
    }
})

response = agent.run("What's the weather in New York and upload it to Dropbox?")
print(response)

Security-native team focused on safe AI adoption.