InstructFenceSafe use and privacy
Keep AI-agent and website-chat testing authorized, scoped, and private.
InstructFence is designed for defensive use. The public site explains the workflow without publishing targets, prompts, operational evidence, or customer information.
Authorization is required
InstructFence does not connect to live agents or websites in its public experience. It provides loopback-only simulations for safe learning and triage; do not treat a simulation as an assessment of a live system.
Private prompts and targets stay private
Prompt text, target configuration, credentials, environment references, runs, findings, evidence, reports, settings, and account data belong in the authenticated console, not public pages.
Local simulations are not production assessments
Safe and intentionally vulnerable simulations are loopback-only development fixtures. A simulated finding demonstrates triage and remediation workflow; it is not evidence about a production system.
Findings follow the selected use case
Findings reflect the selected use case, test pack, and simulation scope. Evidence is redacted for safe review, and reports preserve the context needed to understand the result.
Results have limits
Evaluation results help prioritize defensive improvements and retesting; they do not certify security, compliance, or absence of risk.
Frequently asked questions
Are customer targets and reports public?
No. Operational target, run, finding, evidence, and report data belongs only in the authenticated console.
What does the product not claim?
It does not claim that a test result guarantees a system is secure or compliant.
What data belongs on the public site?
Only reviewed product, workflow, documentation, and safe-use information intended for public audiences.