Skip to main content
Enoki tests AI agents for security and safety vulnerabilities, guards them at runtime, and records what was tested. It connects to your stack over MCP and a REST API. These docs cover the testing workflow: connect an agent, run security and safety assessments against it, then read and remediate the findings. Drive it from your own coding agent over MCP, or from CI/CD over the REST API.

Quickstart

Go from zero to your first assessment in a few minutes.

What we test for

The security and safety coverage, mapped to OWASP and MITRE ATLAS.

Connect your agent

Wire your MCP client to Enoki and register your first target.

MCP tools

The tools your coding agent drives — configure, run, read, remediate.