Sunset Net is a decision-support tool, not an autonomous network administrator. The system generates configuration proposals. It does not execute them without your explicit, deliberate approval — and that distinction is fundamental.
Every command surfaced by Sunset Net must be reviewed, verified, and manually approved by a qualified human operator before it is sent to any device. You are the proximate cause of any change applied to your infrastructure. Sunset Net is the instrument; you are the operator.
I understand that Sunset Net is a decision-support tool, not an autonomous administrator. I am responsible for reviewing, verifying, and manually approving every command before it is executed on any device.
Sunset Net uses Large Language Models (LLMs) — specifically Anthropic's Claude — to generate network configuration syntax. LLMs can and do produce incorrect, non-existent, deprecated, or destructive commands. This is a known and documented limitation of the technology, not a bug unique to this product.
Generated configuration may contain syntax that is invalid for your specific platform, firmware version, or topology. It may reference interfaces, route-maps, or policy objects that do not exist in your environment. It may achieve an output that differs from your stated intent.
Sunset Net does not guarantee that any generated output will: compile without errors, apply successfully, produce the intended network state, or be free of side effects on adjacent protocols or devices.
I acknowledge that Large Language Models can generate incorrect, non-existent, or destructive syntax. Sunset Net does not guarantee that the generated configuration will achieve the intended result or be compatible with my specific firmware version, platform, or topology.
Sunset Net is currently scoped for use in lab, development, and non-production environments — including GNS3, EVE-NG, VIRL/CML, and dedicated test infrastructure. This is not a contractual courtesy; it is a technical reality.
The Safety Engine has been validated in controlled lab environments. It has not been subjected to the full range of edge cases, platform-specific quirks, and failure modes present in live production networks. Applying AI-generated configuration to production infrastructure is outside the intended use of this product.
If you choose to use Sunset Net on production infrastructure, you do so entirely at your own risk. Sunset Net assumes no liability for production outages, packet loss, routing instability, hardware damage, or service degradation resulting from such use.
I agree to use Sunset Net only in lab, development, or non-production environments (e.g., GNS3, EVE-NG). I assume all risk if I choose to apply this tool to production infrastructure.
To the maximum extent permitted by applicable law, Sunset Net's total aggregate liability to you or any third party for any claims arising out of or related to your use of the platform — including network outages, data loss, configuration errors, hardware damage, service disruption, or business interruption — shall be limited as follows:
For users on the free trial: $0 USD. Sunset Net is provided free of charge during the trial period. No consideration has been exchanged. Accordingly, no damages are recoverable.
In no event shall Sunset Net, its founders, employees, or affiliates be liable for indirect, incidental, special, consequential, or punitive damages, including loss of revenue, loss of data, or cost of cover, even if advised of the possibility of such damages.
To the maximum extent permitted by law, Sunset Net's total liability for any network outage, data loss, or hardware damage arising from my use of the platform is limited to $0 for free-trial users. I waive all claims beyond this amount.
Sunset Net processes your natural-language prompts and the outputs of show commands by sending them to Anthropic's Claude API for inference. This means your configuration metadata — including device prompts, topology descriptions, routing intent, and show command output — leaves your local environment and is transmitted to Anthropic's servers.
Anthropic's data handling is governed by their API usage policies and privacy documentation, which you should review independently. Sunset Net does not control how Anthropic stores, processes, or retains data submitted via their API.
You are solely responsible for ensuring that no sensitive personally identifiable information (PII), unencrypted credentials, authentication tokens, SNMP community strings, or proprietary customer data is included in any prompt or show command output submitted to the platform.
I understand that my configuration metadata — including prompts and show command outputs — will be transmitted to third-party AI providers (Anthropic) for processing. I am responsible for ensuring no sensitive PII, credentials, or regulated data is submitted.
Every approval action taken within Sunset Net is recorded in a tamper-evident audit log. When you click Approve on a generated configuration, the following event is written:
{ "timestamp": "2026-02-19T14:22:37Z", "user_id": "eng_01", "device": "core-router-01", "command": "router ospf 1 / router-id 10.0.0.1 / ...", "approved": true, "safety_mode": "interactive", "snapshot_id": "snap_20260219_142237" }
This log constitutes a record of your informed consent to each individual command. In the event of a dispute, Sunset Net reserves the right to present audit log evidence demonstrating that you personally reviewed and approved each configuration change at a specific date and time.
By using the platform, you consent to the creation and retention of these audit records.
On first launch of the Sunset Net application, you will be required to complete the following Safety Check before accessing any functionality. This is not optional and cannot be bypassed.
| Step | Action Required | Legal Effect |
|---|---|---|
| 01 | "I am the human-in-the-loop. I will review every command." | Confirms acceptance of HITL mandate and proximate-cause doctrine. |
| 02 | "I understand that LLMs can produce incorrect or destructive syntax." | Confirms awareness of hallucination risk and waiver of accuracy warranty. |
| 03 | "I am using this tool in a lab or non-production environment." | Confirms scope-of-use agreement and assumption of production risk. |
| 04 | ACTIVATE SAFETYGUARD |
Activates audit logging, snapshot system, and approval workflow. Session begins. |
These terms are governed by applicable law in the jurisdiction where Sunset Net operates. Sunset Net reserves the right to update these terms at any time. Continued use of the platform following notice of updates constitutes acceptance of the revised terms.
If any provision of these terms is found to be unenforceable by a court of competent jurisdiction, the remaining provisions will continue in full force and effect.
These terms represent the entire agreement between you and Sunset Net with respect to liability, safety, and use of the platform, and supersede all prior understandings.
Note: This document is provided for informational purposes. It does not constitute legal advice. Sunset Net recommends consulting qualified legal counsel before deploying AI-powered tooling in regulated or enterprise environments.