Documented relationship
A harness test is connected to a requirement in the crosswalk. This is a research mapping, not a result for any target.
Use the Agent Security Harness to identify relevant tests, run them from a pinned revision, and retain reproducible artifacts for independent review. This reference does not certify a system or determine compliance.
Independent research reference. Not affiliated with or endorsed by AIUC-1.
Use this guide to distinguish a documented requirement relationship from a recorded test run and from independent review. Those are different states of evidence. Treating them as interchangeable is how a useful crosswalk turns into an unsupported readiness claim.
Repository snapshot
Claims model
A harness test is connected to a requirement in the crosswalk. This is a research mapping, not a result for any target.
A stated target and pinned revision were run with retained inputs, environment details, timestamps, and output.
A qualified party assessed the retained evidence and its applicability to the target and relevant standard version.
Practical workflow
Identify the requirement and the applicable AIUC-1 version.
Select mapped tests and define the authorized target and scope.
Record the harness revision, dependencies, and invocation.
Keep inputs, outputs, environment details, and null or negative results.
Seek independent assessment before making a third-party or conformity claim.
Evidence taxonomy
A lower evidence class does not support language reserved for a higher one. A mapping alone is E1-level material, not evidence that a test passed or a control enforced a policy.
Primary artifacts
Requirement-to-test relationships and current claim boundaries.
Module-by-module inventory and the reproducible count command.
Installation and bounded execution guidance.
Permitted language and the evidence each claim class requires.
Versioned releases for pinning and reproducibility.
Known changes, pending scope decisions, and citation limits.
This is a noncommercial research reference. The harness and crosswalk can support evidence collection relevant to a readiness review. They do not certify a system, determine compliance, or substitute for independent assessment of the specific target and standard version.