Fixture sandbox

Inspect the policy model without touching the open internet.

Choose a frozen scenario, change the intended use and inspect both decision stages. Every source signal and entitlement is synthetic or caller-attested fixture data—not a live check, verified contract, license determination or legal conclusion.

Decision boundary: AgentGuard reports an operational result from your configured policy and the evidence supplied to it. It does not grant a license, determine legal rights, or replace legal review.

Fixture library

8 scenarios

01Research source allowedAcquisition and AI-input signals align with the tenant policy.Proceed02Crawler path deniedThe acquisition rule blocks the path before response collection.Deny03Training use deniedFetching is allowed, but the declared model-training use is not.Deny04License availableThe publisher exposes terms, but no matching entitlement is present.License05Entitlement matchedA synthetic contract fixture matches the use subject to attribution and retention.Conditional06Signals conflictDomain-level and response-level evidence disagree, so automation stops.Review07Required evidence staleThe customer policy requires current evidence before extraction.Review08Response changes resultStage one allows acquisition; a response header changes the final use decision.Deny

Display URLhttps://fixture.example/research/briefThis URL is descriptive fixture data. It is never requested.

Declared purposeRetentionOutputMode

Proceed

Stage 1Acquisition

Proceed

REP_PATH_ALLOWED

Stage 2Response use

Proceed

USE_SIGNAL_ALLOWED

Synthetic evidence claims3

FRESH

Path permitted

User-agent and resource path are not disallowed.

robots.txt

FRESH

AI input permitted

ai-input=yes; ai-train=no

content-usage

FRESH

Tenant policy matched

External research is allowed with a 30-day retention limit.

tenant-policy

Sample decision recorddec_fixture_clear-research

Policy revisionpolicy_demo_018

Evaluatorfixture-0.1.0

Valid until9/4/2026, 1:00:00 PM