blockconvey-monitor 0.3.0

Production AI agent observability for regulated industries

blockconvey-monitor

Production AI agent observability for regulated industries.

Install

pip install blockconvey-monitor

Quick start

from blockconvey import monitor

prism = monitor(
    api_key="pt_your-key",
    project_id="your-project-id"
)

# Send a trace
prism.trace(
    input_messages=[{"role": "user", "content": user_input}],
    output_message=agent_response,
    model="gpt-4o",
    latency_ms=1200,
    agent_name="Customer Service Agent"
)

With decorator

from blockconvey import monitor, traced

prism = monitor()

@traced(prism, agent_name="LoanAdvisor", model="claude-sonnet-4-6")
def ask_agent(user_message: str) -> str:
    return your_llm_call(user_message)

Async

from blockconvey import async_monitor

prism = async_monitor()

await prism.trace(
    input_messages=[{"role": "user", "content": user_input}],
    output_message=agent_response,
    model="gpt-4o",
)

Environment variables

BLOCKCONVEY_API_KEY=pt_your-key
BLOCKCONVEY_PROJECT_ID=your-project-id

Integrations

LangChain / LangGraph

pip install blockconvey-monitor[langchain]

from blockconvey.integrations.langchain import BlockConveyCallbackHandler
from langchain_anthropic import ChatAnthropic

handler = BlockConveyCallbackHandler(
    api_key="pt_your-key",
    project_id="your-project-id",
    agent_name="MyAgent"
)
llm = ChatAnthropic(model="claude-sonnet-4-6", callbacks=[handler])
# All llm.invoke() calls are now automatically traced

# LangGraph
from blockconvey.integrations.langgraph import BlockConveyCallbackHandler

handler = BlockConveyCallbackHandler(agent_name="LoanProcessingGraph")
graph.invoke(state, config={"callbacks": [handler]})

OpenAI

pip install blockconvey-monitor[openai]

import openai
from blockconvey import monitor
from blockconvey.integrations.openai import wrap_openai

prism = monitor()
client = wrap_openai(openai.OpenAI(), prism)

# All client.chat.completions.create() calls are now auto-traced
response = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Explain model risk management."}],
)

Anthropic

pip install blockconvey-monitor[anthropic]

import anthropic
from blockconvey import monitor
from blockconvey.integrations.anthropic import wrap_anthropic

prism = monitor()
client = wrap_anthropic(anthropic.Anthropic(), prism)

# All client.messages.create() calls are now auto-traced
response = client.messages.create(
    model="claude-sonnet-4-6",
    max_tokens=1024,
    messages=[{"role": "user", "content": "What is SR 11-7?"}],
)

Install all integrations

pip install blockconvey-monitor[all]

Docs

https://docs.blockconvey.com