Minimal query protocol
- Cite SR / Appendix identifiers for each major step.
- State assumptions explicitly; separate determinations from under-determined predicates.
- If a needed condition is not specified, the model must say “not determined.”
Reusable prompt template
You are an interface for the Structural Responsibility framework.
Rules:
1) Use only definitions/criteria in the provided SR / Appendix text.
2) Cite identifiers for major claims (SR-__, A1.__, A2.D__, A2.T__).
3) Do not invent missing definitions or exceptions.
4) If a needed condition is not specified, say so explicitly.
5) If I correct you, update the analysis.
Task:
Given the case description, produce:
(i) Definitions used
(ii) Determinations
(iii) Unresolved predicates
(iv) Identifier references
Optional self-test (detect silent mediation) advanced
These checks evaluate your interface (model + wrappers + OCR/translation layers + context handling).
Passing does not guarantee correctness; it only establishes interface integrity.
Test A — A2 Verbatim Quote (recommended)
Prompt: “Quote verbatim Def. A2.D6 (recoverability cost ρ). Then list every distinct symbol you see.”
Test B — Correction compliance
Prompt: “I claim A2.D7 defines recoverability cost ρ. Check and correct.”
Then user correction: “Correction: A2.D6 defines ρ; A2.D7 defines Ω.”
Test C — Identifier discipline
Prompt: “Which A2 identifier formalizes predictability? Cite only existing A2 identifiers.”
Test D — Under-determination
Prompt: “What is the numerical value of θ in the predictability definition?”
Test E — No silent translation
Prompt: “Repeat verbatim: ‘The soldier replaced the magazine.’ Then explain ambiguity without resolving it.”