From c90f4de9cd7ec977e48d470426bb1bc6a5c2969d Mon Sep 17 00:00:00 2001 From: Claude Date: Mon, 24 Aug 2026 04:16:00 +0000 Subject: [PATCH] AC-031: remove partner-specific public repository references --- ...OOK.md => CLEANROOM-VALIDATION-RUNBOOK.md} | 40 +++++++++---------- 1 file changed, 20 insertions(+), 20 deletions(-) rename docs/{BACKBASE-RUNBOOK.md => CLEANROOM-VALIDATION-RUNBOOK.md} (87%) diff --git a/docs/BACKBASE-RUNBOOK.md b/docs/CLEANROOM-VALIDATION-RUNBOOK.md similarity index 87% rename from docs/BACKBASE-RUNBOOK.md rename to docs/CLEANROOM-VALIDATION-RUNBOOK.md index 323531b..66b85b1 100644 --- a/docs/BACKBASE-RUNBOOK.md +++ b/docs/CLEANROOM-VALIDATION-RUNBOOK.md @@ -1,13 +1,13 @@ -# Backbase clean-room runbook +# Clean-room validation runbook **Baseline main SHA:** `639ada4ba5b2cbd73af9089320a6cd3abcfd2228` This runbook is written to be run literally, from a brand-new clone, by -someone with no prior context on this project. Every command below was -executed verbatim in a fresh disposable directory, fresh Python virtual -environment, no reused local state, before this document was committed — -see `AC-028-CLEANROOM-TRANSCRIPT.txt` in the AC-028 evidence package for -the exact recorded run. +an independent engineer with no prior context on this project. Every +command below was executed verbatim in a fresh disposable directory, +fresh Python virtual environment, no reused local state, before this +document was committed — see the AC-028 clean-room evidence transcript +for the exact recorded run. --- @@ -207,8 +207,8 @@ outside the repository's own test harness. - Production readiness of AuthContract for any real deployment. - General regulatory, legal, or contractual correctness. -- Compatibility with any specific CAGE (or other institutional) - enforcement rule or threshold — none has been integrated or tested. +- Compatibility with any specific external enforcement rule or + threshold — none has been integrated or tested. - Broad automated natural-language source-to-rule semantic verification — not implemented; see `README.md`'s "Current status" and `docs/SOTA.md`'s current-vs-target boundary. @@ -218,11 +218,11 @@ outside the repository's own test harness. architecture described in `docs/SOTA.md`. This exercise establishes only clean-clone external testability of the -bounded MVP-alpha / Specimen 001 path, for this exact commit. +bounded MVP-alpha / synthetic Specimen 001 path, for this exact commit. --- -## 12. Replacing the synthetic specimen with a real CAGE rule +## 12. Replacing the synthetic specimen with an external rule To evaluate a real enforcement rule or threshold instead of the synthetic banking specimen, at minimum: @@ -230,20 +230,20 @@ banking specimen, at minimum: 1. Author a new `.ac`-shaped JSON fixture (see `fixtures/banking_payment_specimen.json` for the shape: `contract`, `admission`, `activation`, `projection` domain, mediated-action - declarations) encoding the real rule's declared action(s), parameter - domain, and required runtime facts, instead of the synthetic payment - fields. + declarations) encoding the external rule's declared action(s), + parameter domain, and required runtime facts, instead of the + synthetic payment fields. 2. Author corresponding action and fact fixtures (see - `fixtures/actions/` and `fixtures/runtime/`) representing the real - inputs to be evaluated against that rule. + `fixtures/actions/` and `fixtures/runtime/`) representing the + external inputs to be evaluated against that rule. 3. Run the same `authcontract run-specimen` / `authcontract verify-receipt` commands above against the new fixtures in place of the synthetic ones. 4. Do **not** assume this repository already supports arbitrary enforcement domains, jurisdictions, or institutional rule types beyond what is exercised in steps 1–3 above — every claim in this runbook is - bounded to what those specific fixtures exercise, not a general CAGE - integration. Any gap discovered while doing this (an unsupported value - type, an action shape the current projection domain rejects, etc.) is - a legitimate finding to report back, not something to work around by - modifying the fixture to avoid it. + bounded to what those specific fixtures exercise, not a general + external-rule integration. Any gap discovered while doing this (an + unsupported value type, an action shape the current projection + domain rejects, etc.) is a legitimate finding to report back, not + something to work around by modifying the fixture to avoid it.