Integration

LangChain & LangGraph

Wrap any tool your agent invokes so governance runs before dispatch — not after observability logs the damage.

Install

npm install @datavibe.cc/sdk
pip install datavibe

TypeScript

import { DataVibeClient } from "@datavibe.cc/sdk";
import { wrapTool } from "@datavibe.cc/sdk/integrations";

const dv = new DataVibeClient({ apiKey: process.env.DATAVIBE_API_KEY! });

const guardedSend = wrapTool(dv, sendEmailTool, {
  contentType: "agent_action",
  sourceModel: "claude-sonnet-4-6",
  onBlocked: (v) => `Blocked: ${v.violations[0]?.rule}`,
  onReviewRequired: (v) => `Review: ${v.review_url}`,
});

Python

from datavibe import DataVibeClient
from datavibe.integrations.langchain import wrap_tool

dv = DataVibeClient(api_key="dv_live_…")
guarded = wrap_tool(dv, send_email_tool, content_type="agent_action")

Content types

Use agent_action for tool calls and business actions, tool_call for raw tool payloads, email for outbound messages, and api_response for generated API bodies.

Agentic AI governance · SDK guide · Live simulator · Getting started