vequil 0.1.0

Reliability and observability SDK for AI agent operators

Vequil Python SDK

Reliability and observability for AI agent operators.

pip install vequil

Quickstart

from vequil import VequilClient

vq = VequilClient(workspace_key="vk_ws_...")

# Log any agent event
vq.ingest(
    source="openclaw",
    event_type="tool_call",
    event_status="success",
    agent_id="main-agent",
    tool_name="bash",
    cost_usd=0.012,
)

# Auto-log success or failure with a context manager
with vq.watch(source="openclaw", agent_id="main", tool_name="bash"):
    run_the_thing()

Framework integrations

LangChain

from vequil.integrations.langchain import VequilCallbackHandler
handler = VequilCallbackHandler(vq, agent_id="my-chain")
chain.invoke({"input": "..."}, config={"callbacks": [handler]})

CrewAI

from vequil.integrations.crewai import VequilCrewObserver
observer = VequilCrewObserver(vq, agent_id="my-crew")
observer.on_task_start("research", agent_name="researcher")

AutoGen

from vequil.integrations.autogen import patch_agent
patch_agent(assistant, vq, session_id="session-1")

LangGraph

from vequil.integrations.langgraph import vequil_node

@vequil_node(vq, agent_id="my-graph", tool_name="search")
def search_node(state):
    ...

Environment variables

Variable	Description
VEQUIL_WORKSPACE_KEY	Your workspace ingest key
VEQUIL_URL	Custom Vequil instance URL (default: https://vequil.com)

Get your workspace key at vequil.com/quickstart.