# Governed Admissibility Tester Guide Generated: `2026-06-14` ## Purpose This guide explains how testers in different disciplines should understand governed admissibility. The test is not asking whether the output is interesting, fluent, useful, or impressive. The test asks: ```text What is this output allowed to become? ``` ## Why this applies across disciplines Governed admissibility applies whenever an output, claim, instruction, artifact, or proposed transition could become a consequential next state. Examples: ```text text → public claim summary → policy posture solver output → proof claim agent plan → tool action code suggestion → deployment paper citation → evidence support health explanation → care decision financial summary → allocation decision lesson output → assessment or curriculum ``` ## What testers should classify For each test object, identify: ```text 1. Declared intent 2. Domain context 3. Authority source 4. Evidence posture 5. Replayability 6. Consequence level 7. Claim limit 8. Allowed next state ``` ## Core admissibility decisions ```text ALLOW ALLOW_AS_NOTE ALLOW_WITH_POSTURE REQUIRE_REVIEW REQUIRE_TEST REQUIRE_RECEIPT REQUIRE_REPLAY HOLD FAIL_CLOSED NOT_ADMISSIBLE ``` ## Discipline examples ### AI / LLM systems A model response may be harmless as a draft but not admissible as a public claim, instruction, or autonomous action. Test object examples: ```text model response tool call agent plan generated artifact ``` Primary question: ```text Can this output become a claim, tool action, instruction, or only a draft? ``` ### Mathematics / formal methods A derivation or solver artifact may be useful as a research note without being admissible as proof. Test object examples: ```text lemma proof attempt derivation counterexample solver artifact ``` Primary question: ```text Can this artifact become a proof claim, or must it remain a bounded research note? ``` ### Software engineering A code suggestion becomes more consequential when it is merged, deployed, scheduled, or granted credentials. Test object examples: ```text pull request script workflow configuration change deployment action ``` Primary question: ```text Is this change authorized, replayable, and safe to execute now? ``` ### Cybersecurity / identity Identity and access actions require strong authority boundaries because mistakes can create invisible trust failures. Test object examples: ```text token use identity assertion permission change secret handling access request ``` Primary question: ```text Is this access or identity transition authorized, evidenced, and consequence-bounded? ``` ### Legal / policy Legal or policy text can be useful context, but it must not become advice, compliance posture, or institutional position without review. Primary question: ```text What claim is allowed, and what domain review is required before stronger posture? ``` ### Medicine / health Health-related output can cross from explanation into advice or care direction. Primary question: ```text Is this only general information, or is it being treated as care guidance requiring domain review? ``` ### Finance / markets Financial outputs can influence risk, payment, investment, or allocation decisions. Primary question: ```text Is this a contextual note, a risk claim, or an action recommendation requiring review? ``` ### Education / learning Educational outputs can affect learning paths, assessments, and claims about competence. Primary question: ```text Is this output supporting learning, replacing judgment, or creating an assessment claim? ``` ### Science / research Research outputs require source posture, reproducibility posture, and claim limits. Primary question: ```text What can be claimed from this evidence, and what must remain hypothesis or research note? ``` ### Engineering / infrastructure Engineering outputs may affect physical systems, uptime, safety, or maintenance. Primary question: ```text Is this recommendation authorized, reviewed, and safe to move into operation? ``` ### Robotics / autonomy Autonomous instructions can become physical-world actions. Primary question: ```text Should this plan execute, hold, require review, or fail closed? ``` ### Journalism / public information Public information outputs can become claims, narratives, or timelines. Primary question: ```text Is the source posture strong enough for publication, or only for draft/research use? ``` ### Government / civic systems Civic outputs can affect eligibility, services, authority, or public trust. Primary question: ```text Does this action have authority, evidence, replayability, and consequence review? ``` ### Organizational governance Governance decisions require authority, quorum, evidence, and refusal consequences. Primary question: ```text Is the decision valid now, under current authority and evidence, at commit time? ``` ### Archives / records / history Historical claims require source posture and provenance. Primary question: ```text Can this become a timeline claim, or must it remain a source note? ``` ## Tester output shape A tester should be able to return: ```text Discipline: Test object: Declared intent: Authority source: Evidence posture: Replay posture: Consequence level: Decision: Allowed next state: Required follow-up: Claim limit: ``` ## Boundary This guide explains where governed admissibility testing applies. It does not replace domain expertise, professional review, formal proof, medical review, legal review, financial review, engineering sign-off, or civic authority.