[Task]: Secret-types boundary guard: keep the pointer accurate, keep claims out (successor to #32) #39

Category

Chore / maintenance

Description

Successor to #32 (closed as delivered). #32's model/boundary deliverable is
recorded in docs/secret-types.adoc; the specification itself and all formal
follow-on work live in hyperpolymath/secret-types (D153/D154; owner
confirmation 2026-10-04; ADR docs/decisions/0004-standalone-repo-and-scope.adoc,
PR #4 d1122d8).

This is a guard, not a re-run. Do not re-open the specification here. Model
changes, lattice choices, the noninterference theorem, declassification policy,
and any Secret module belong to hyperpolymath/secret-types.

Scope — three items only

1. Sync the pointer when the standalone spec's status changes. When its
   "Review checklist before implementation" is completed item by item, or the
   spec is accepted, renamed, or retired, update the status and boundary lines
   in docs/secret-types.adoc, .machine_readable/descriptiles/STATE.a2ml,
   .machine_readable/descriptiles/META.a2ml (ADR-007), and
   0-AI-MANIFEST.a2ml, citing the exact secret-types revision. Keep the
   2026-10-05 transition review as history: add a dated note rather than
   rewriting it.
2. Keep Secret/IFC claims out of this repository. No Secret type, IFC
   calculus, declassification rule, noninterference theorem, or
   confidentiality/leakage guarantee may attach to Access.Preorder,
   BeliefModality (missing reflect), Warrant, or EchoBridge.Grade here.
   If one appears, that is a defect — comment with the file and line, or open a
   focused bug.
3. Record, don't edit, the downstream items (owned elsewhere): the family
   map is nextgen-typing#118 / docs/TYPE-CONNECTIONS.adoc (no secret-types
   row as of 2026-10-05); the feeder text is valence-shell
docs/THEORY-FEED.adoc under #210 (still says un-minted and "mint it … or
   fold it into epistemic-types#32"; valence-shell#92 remains unselected);
   the mint/scaffolding is secret-types#2.

Close condition: close when item 1 is done for the standalone spec's current
status, or when the owner judges the pointer final. This issue is not closed by
the mere existence of hyperpolymath/secret-types, and it is not a reason to
re-open #32.

Rationale

History: #32 (D154 comment 2026-09-30), #29 (closed 2026-09-27),
docs/secret-types.adoc (PR #33 d97ecaab; boundary record PR #37 eb810d4;
2026-10-05 transition review), secret-types PR #4 (d1122d8).

Before submitting

• I searched existing issues to avoid duplicates.