[masarray]

Stacked follow-up to PR #150.

Goal

Give Open SCL and Discovery the same engine-owned runtime bridge from canonical static coverage to actual live RCB evidence without forcing a full IED discovery.

PR #150 correctly refuses to promote SCL/canonical structural inventory into operational live evidence. The missing next step is obtaining just enough live evidence, cheaply and deterministically.

Runtime strategy

For selected static coverage only:

1. extract the exact MMS domains and required RCB families (BRCB/URCB) from canonical coverage;
2. enumerate NamedVariable only for those domains;
3. parse only real live RCB candidates from that live directory evidence;
4. discard unrelated RCB classes/domains;
5. read whole-RCB structure only for those required-family candidates;
6. if DatSet was not exposed by the structure, read only its exact DatSet attribute;
7. return a LiveMmsObserved inventory;
8. feed that into PR P6 #144: source-aware canonical static acquisition preflight #150 exact canonical -> live target resolution and PR P4 #144: targeted RCB hot path and association-scoped static DataSet evidence #147 targeted availability.

No FC-point discovery, DataSet-list discovery, unrelated-domain enumeration, dynamic DataSet mutation, sibling-name fabrication or full-discovery fallback is introduced.

Performance boundaries

• domains are de-duplicated locally before network work;
• independent domain GetNameList chains reuse negotiated outstandingCalling limits;
• domain and RCB-candidate budgets are hard fail-closed guards;
• budget overflow never probes an arbitrary partial subset;
• RCB enrichment is limited to exact required family candidates;
• no fixed sleeps or retry loops.

New engine entry point

PrepareCanonicalStaticAcquisitionSmartAsync(...)

This is the intended thin-consumer contract:

canonical model + selected signals
        ↓ local
configured static coverage
        ↓ bounded network
targeted live RCB reconciliation
        ↓
exact live RCB target
        ↓
targeted availability

Full discovery is not implied.

Tests

• multiple selected DataSets in the same LD collapse to one domain enumeration;
• BRCB/URCB requirements are merged per domain;
• fallback domain extraction remains exact;
• domain-budget overflow authorizes zero network work;
• no-static-coverage path performs zero network work even on a disconnected session.

Teamwork

This PR remains engine-only. It does not modify ARSAS PR #425 or its engine lock. Once software-gated, ARSAS issue #446 can consume this contract on a stacked child of #425 instead of adding more application-side protocol logic.

Software gate

Combined P0-P6.1 exact integration head 9c5292570f55dd81be1b3a6b56f937e5ba1ed276 passed .NET CI #759 end-to-end:

• source/provenance verification;
• restore;
• Release build;
• complete test suite;
• package;
• artifact upload;
• diagnostics upload.

This is CodeVerified only. It is not yet a production/main merge or PhysicalVerified claim.

[masarray]

Staging reconciliation 2026-10-22: this old standalone PR's intended functional scope exists in the newer staging implementation, but the old branch still diverges from the newer code. Do NOT cherry-pick the historic file snapshots. Original HEAD c674518f9458846ce0b4a77a57bcd2e911579e75 is not claimed an ancestor of staging.

Code and test evidence: Staging has PrepareCanonicalStaticAcquisitionSmartAsync, MmsCanonicalStaticLiveReconciliationPlanner, bounded per-domain reconciliation and tests. MmsCanonicalStaticLiveReconciliationPlannerTests.cs is byte-identical between original branch and staging, while staged runtime extends it with more recent candidate pagination.

This source scope passed Windows PR #192 synthetic merge CI #39782663272: 1,262/1,262 tests, zero warnings/errors, exact staging SHA 44ac7554f43f2ba06763f3895242cb569b2e6a1c. Closing old PR as functionally superseded (without merge); Git branch/commits retained, and staging→main #192 stays DRAFT pending consumer and physical acceptance. Reopen if a specific missing invariant is documented.