Skip to content
VulnifyVulnifyVulnify

What is Vulnify

Vulnify decides what an AI agent is allowed to do before the action runs. The agent, or the code around it, describes the action. Vulnify returns ALLOW, REVIEW, or BLOCK in real time. A review waits for a person, for example in Slack. Every decision is appended to a hash-chained audit log.

Vulnify is a runtime authorization layer. It does not scan machines or source code for vulnerabilities. It governs what agents and automations can access and do.

The product is in early access. The web app is app.vulnify.io. The API is https://api.vulnify.io.

What Vulnify sees

Section titled “What Vulnify sees”

A check sends action metadata: the agent, the action, the resource, the destination, and how many records are affected. Vulnify is designed to store that metadata, not the contents of your records.

You can optionally send content so the server can scan it for sensitive data. Matches come back as findings. The content itself is not stored. See Risk engine.

What you do with the decision

Section titled “What you do with the decision”
Decision What it means
ALLOW Run the action.
REVIEW Do not run the action. A person has to approve it.
BLOCK Do not run the action.

In monitor mode the event is stored and not enforced. Follow finalDecision, which matches the recorded decision when there is no review. evaluatedDecision is what enforcement would have returned. If Vulnify cannot be reached, the SDKs fail closed by default and return BLOCK. A rejected request, including a body over 200 KB, throws instead.

Two ways to ask

Section titled “Two ways to ask”
  1. Check, then act. Your code calls the API or an SDK, and runs the tool only after ALLOW. This is the path the SDKs implement.
  2. Gateway. The app can register an HTTP API or an MCP server as a route. Callers send the call to Vulnify. Vulnify decides, and forwards it to the upstream only when the decision allows it. The app describes this path as beta. A credential for the upstream is injected only after ALLOW, and the agent never sees it.

Where to go next

Section titled “Where to go next”

Quickstart

Create an API key, install an SDK, and authorize one export. Start here.

Decisions

ALLOW, REVIEW, BLOCK, monitor mode, and fail-closed. Read the model.

Node.js and Python

Clients, guard(), and the framework adapters that ship in the SDKs. Node.js · Python.

API reference

POST /v1/events and the gateway, authenticated with an API key. Open the reference.