Control Zero is in early beta. We ship daily. Feedback: team@controlzero.ai

Integrations

Works with your stack.

AI providers, alert channels, and MCP clients, all governed from one control plane.

// 000

Enforcement Surfaces

One policy, multiple places it fires. Hard enforcement where it can, cooperative enforcement where it cannot, discovery for the places that do not speak policy at all. Browser extension coming soon.

SDK

Hard

Hard enforcement on the developer machine. Every tool call your agent makes runs through the on-device evaluator before executing.

ModesHostedHybridLocal

Gateway

Hard

Hard enforcement in front of the LLM provider URL. Transparent proxy. DLP rules tagged for the gateway scope fire on every request.

ModesHostedHybridLocal

MCP Guard

Cooperative

Cooperative enforcement via three MCP tools (check, test, list) any MCP-capable client can call before invoking other tools.

ModesHostedLocal

Browser Extension

Coming Soon

Paste-boundary enforcement on Claude.ai, ChatGPT, and Gemini web clients. Scans clipboard against the org DLP rules before the paste lands. Coming soon -- not yet published to browser stores.

ModesHostedHybrid

Shadow AI Scout

Discovery

Discovery-only observability for unsanctioned AI tool usage across SaaS catalogs and billing exports. Feeds the audit log with source=scout.

ModesHostedHybrid

// 001

AI Providers

Anthropic

Stable

Claude 4, Claude 3.5, Claude 3. Full streaming, tool use governance, and DLP scanning.

OpenAI

Stable

GPT-4, GPT-4 Turbo, GPT-3.5, o1, o3. Full streaming and tool call support.

Google Gemini

Stable

Gemini Pro, Gemini Ultra, Gemini Flash. Dedicated gateway with streaming, tool call governance, and DLP.

Ollama

Stable

Local models with full policy enforcement. OpenAI-compatible wire format.

DeepSeek

Stable

DeepSeek Chat, DeepSeek Coder. OpenAI-compatible routing with policy enforcement.

MoonshotAI

Stable

Moonshot v1 models. OpenAI-compatible routing with policy enforcement.

HuggingFace TGI

Stable

Text Generation Inference. Streaming support, no tool interception.

// 002

Alert Channels

Slack

Real-time policy violation alerts and daily digest summaries to Slack channels.

Email

Configurable email alerts for critical policy violations and usage thresholds.

Telegram

Instant notifications to Telegram bots for on-the-go monitoring.

Discord

Webhook-based alerts to Discord channels for team visibility.

Webhook

Generic webhook endpoint for custom integrations and event pipelines.

// 003

MCP Server

The Control Zero MCP server gives your AI assistant the ability to manage Control Zero policies via prompt. Install in any MCP-compatible client. To govern what your AI tools DO, see the coding hooks section.

Claude Desktop

Govern MCP tool calls made by Claude Desktop with policy-driven access control.

Cursor

Enforce tool restrictions and audit all MCP operations in Cursor.

Windsurf

Policy enforcement for Windsurf MCP tool calls and file access.

VS Code

Govern MCP tools used by Copilot and other VS Code AI extensions.

Gemini CLI

Enforce governance on MCP tool calls from the Gemini CLI.

// 004

Coding Assistant Hooks

One-command install. Every tool call your coding assistant makes is evaluated against your policy before it executes. Full audit log from day one.

Claude Code

PreToolUse hook intercepts every tool call. Install in one command. Audit log from day one.

controlzero install claude-code

Gemini CLI

preToolCall hook governs shell commands, file writes, web fetches, and MCP tools.

controlzero install gemini-cli

Codex CLI

tool_approval_policy wrapper evaluates every tool call against your policy before execution.

controlzero install codex-cli

Cursor

MCP and built-in tool governance. Allows read-heavy tools, denies dangerous shell operations.

controlzero install cursor

Windsurf

Policy enforcement for Windsurf tool calls and file access operations.

controlzero install windsurf

// 005

Framework Integrations

Drop-in governance for the most popular AI agent frameworks. Wrap your existing code. Policy enforcement and audit logging are automatic.

LangChain

Wrap LangChain agents to govern tool calls, chain executions, and LLM invocations.

View docs →

LangGraph

Policy enforcement for LangGraph stateful agents and multi-step workflows.

View docs →

CrewAI

Govern multi-agent crews with delegation controls and inter-agent communication policies.

View docs →

AutoGen

Guardrails for AutoGen code execution and named tool invocations.

View docs →

OpenAI Agents SDK

Policy enforcement for OpenAI Agents SDK tool calls, handoffs, and guardrails.

View docs →

Pydantic AI

Govern Pydantic AI agent tool calls with type-safe policy enforcement.

View docs →

MCP Servers

Policy-driven access control for any MCP-compatible tool server.

View docs →

// 006

Browser Extension

Coming Soon

DLP overlay for browser-based AI chat applications. Scans pastes before they reach the provider. Deploy across your fleet with zero configuration per device. Currently in development -- not yet published to browser extension stores.

Chrome / Edge MV3

Manifest V3 extension deployed via Chrome Enterprise, Jamf, or Intune. Force-install across your fleet.

AI Chat DLP Overlay

Scans input fields and clipboard pastes on Claude.ai, ChatGPT, Gemini, and Perplexity. Blocks sensitive content before it reaches the provider.

Enterprise Deployment

Zero-touch deployment via MDM. Rules sync from the dashboard. Fail-closed on cold boot. Compliance acknowledgment gate.

// 007

Policy templates

Nine opinionated starting points. controlzero init -t <name> writes a template into your project. Edit from there.

generic

Hello World policy with comments. Start here if you are not sure what you need.

claude-code

Installs as a Claude Code PreToolUse hook. Allow-by-default with opinionated deny rules.

langchain

LangChain / LangGraph agents. Gates Tool calls and denies destructive DB operations.

crewai

Multi-agent crews with delegation. Allows inter-agent comms, denies publishing / outbound messages.

cursor

Cursor editor MCP + built-in tools. Allows read-heavy tools, denies danger-tagged shell.

autogen

AutoGen with code_execution_config guardrails. Denies bare code execution, allows named tools.

mcp

Generic MCP server catalogue with the most common sanctioned tool patterns.

rag

Retrieval-augmented generation: vector lookups, document fetches, answer writes.

cost-cap

Token budget and cost ceilings per-tool and per-provider.

Need a different integration?

We are adding new integrations based on community feedback. Let us know what you need.