LIMEN13

About

An operational security view of AI agents

LIMEN13 is a founder-led security engineering project in its pilot phase.

In plain words. LIMEN13 comes from day-to-day security work: watching alerts, stopping incidents and proving afterwards what happened. It applies the same habits to AI agents.

Founder

Founder background

LIMEN13 was founded on hands-on experience in security operations, cybersecurity engineering, and security automation, including practical work with EDR and SIEM technologies. The project brings an operational security perspective to AI-agent execution, focusing on enforceable controls, incident containment, and verifiable evidence.

Introductions

LIMEN13 is founder-led. Introductions are made directly, in the first pilot discussion.

Origin

Why LIMEN13 exists

AI agents now hold credentials and take actions: they send email, issue refunds and change records. After an incident, a security team asks the same questions every time. What acted? With whose authority? What stopped it? Where is the proof? LIMEN13 was built to make those questions answerable for agents.

Limen is the Latin word for threshold: the doorway every agent action has to cross.

Engineering philosophy

How it is built

Code decides, not the model

Deterministic permissions come first. Detection is a second line.

Deny by default

An agent cannot misuse a capability it was never given.

Evidence over assurance

A claim on this site points to a test, a record or a stated limit.

State the limits

We publish what LIMEN13 does not stop, including the attack cases it fails.

Security principles

Principles we hold to

  • Standard cryptography from a maintained library: AES-256-GCM and HMAC-SHA-256. No home-made algorithms.
  • Secrets are shown once, then kept only as digests or encrypted.
  • No telemetry: the software does not contact us.
  • Fail closed: without its signing key, or with corrupted evidence, the engine does not start.
  • No customer logos, certifications or badges until they are real.

What to expect

Contact and support boundaries

  • One address for business, pilots and security reports: contact@limen13.com.
  • Vulnerability reports are handled as described in the trust center.
  • No managed service, on-call response or production support commitment is offered in the pilot phase.
  • The terms of a pilot are agreed in writing before it starts.

Status

Where the project stands

Stage
Pilot program, release 0.10.1.
Deployments
Assisted pilots on non-production systems.
Independent testing
Not yet done. Required before production.
Certifications
None.
Contact
contact@limen13.com