Pilot and pricing
Test LIMEN13 on one of your own agents
An assisted pilot on a non-production system. You see the controls work on your agent, and you keep the results.
In plain words. A pilot is a guided trial. We help you install LIMEN13 next to one of your AI agents on a test system, run seven tests together, and you keep the results. Prices are given on request.
Process
Seven steps
- Initial technical discussionYour agents, their tools, and what worries you.
- Scope and security prerequisitesOne agent, one workflow, one non-production system; who approves; where it runs.
- Non-production deploymentWe help you install LIMEN13 on your server, with Docker or systemd.
- Agent integrationPoint the agent's model SDK, REST tools or MCP client at LIMEN13, and move its credentials into LIMEN13.
- Controlled enforcement testingObservation mode first, then the seven tests below.
- Evidence validationExport the record and verify it yourself.
- Findings and next stepsA written summary: what worked, what did not, and what production would require.
Tests
The seven tests we run together
- A prompt injection hidden in a malicious email
- An unauthorized API or MCP tool action
- A direct call that tries to go around LIMEN13
- An action that needs a person's approval
- Disabling an agent in the middle of a task
- Exporting and verifying the evidence yourself
- Measuring the added delay, and what happens if LIMEN13 stops
Deliverables
What you keep
- A working LIMEN13 on your non-production server
- The policies written for your agent
- The result of each of the seven tests
- Your signed evidence record and its verification output
- A monthly-report draft generated from your pilot data
- A written summary of findings and limits
Prerequisites
What we need from you
- An agent that already runs, or is in development
- A non-production copy of one business system
- A Linux server or virtual machine you control, with Docker or Python 3.12 or newer
- A person who can approve actions, and a technical contact
Who it is for
Who a pilot suits
- A team that already runs an AI agent, or is about to
- A security provider that wants to evaluate it for its customers
- One agent, one workflow, one non-production system
- A named technical contact and a named approver on your side
Responsibilities
Who does what during the pilot
| Topic | You | LIMEN13 |
|---|---|---|
| Environment | Provide a Linux server or virtual machine and a non-production copy of one business system. | Provide the release and the installation steps. |
| Deployment | Run the installation on your server and keep the keys. | Assist remotely during installation. |
| Integration | Point the agent at LIMEN13 and move its credentials into it. | Help write the first policy for the agent. |
| Testing | Run the seven tests with us and judge the results. | Guide the tests and explain each decision. |
| Security boundary | Keep production systems and real customer data out of the pilot. | No access to your server is needed, and nothing is sent to us. |
| Data | Your evidence and configuration stay on your server. | We keep only the emails and notes you send us. |
Terms
Period, criteria, support and exit
- Evaluation period
- Agreed in writing before the pilot starts. No standard length is published.
- Evaluation criteria
- The seven tests, the added delay on your setup, and whether your normal work still completes.
- Support during the pilot
- By email and scheduled calls. No response-time commitment is published.
- Supported environments
- Linux, with Docker or Python 3.12 or newer. Agents that reach models through OpenAI-compatible or Anthropic APIs, and tools through REST or MCP.
- Exit and cleanup
- Stop the engine, delete its folder and its keys, and restore the agent's own credentials. Nothing remains on our side except your emails, which we delete on request.
- Price
- Quoted on request.
Pricing
Pricing
| Offer | Price |
|---|---|
| Pilot | Contact us |
| Production licence | Custom quote |
| MSSP licence | Contact us |
No prices are published yet
Licensing may depend on dimensions such as the number of protected agents, deployment instances, customer environments, support level or usage volume. These are possible models, not final commercial terms.
Contact
How to ask
- Write to contact@limen13.comA plain email is enough.
- Tell us about one agentWhat it does, the tools and models it uses, and where it runs.
- We reply by emailTo arrange the first technical discussion.
Privacy
This site has no form and no tracking. If you write to us, we use your email only to answer you. Privacy policy.
Start with one agent.
One agent, one workflow, one non-production system.
Assisted pilot program. No production commitment is implied.