fix: harden store API follow-up

.claude/commands/audit-code.md

@@ -148,6 +148,8 @@ Compare current implementation against latest platform APIs:
 
 **Internal Rules Compliance:**
 
+- [ ] Canonical KISS/SSOT rules in `03-coding-style.md` pass
+
 packages/apple (Swift):
 
 - [ ] iOS-specific functions end with `IOS` suffix
@@ -192,7 +194,8 @@ After identifying issues:
 
 1. Read the relevant knowledge file for the rule
 2. Read the violating code file
-3. Fix the code to comply with the rule
+3. Fix the code to comply with the rule, simplifying or consolidating before
+   adding another layer
 4. For missing features: add to roadmap or implement
 
 ### 7. Update Documentation

.claude/commands/review-pr.md

@@ -31,6 +31,8 @@ Based on changed files, run these checks BEFORE committing:
 
 When reviewing, check these project-specific rules:
 
+- **KISS/SSOT**: Enforce the canonical release rules in
+  `knowledge/internal/03-coding-style.md`
 - **iOS functions**: Must end with `IOS` suffix (e.g., `syncIOS`)
 - **Android functions in packages/google**: NO `Android` suffix (it's Android-only)
 - **Generated files**: Do NOT edit `packages/apple/Sources/Models/Types.swift` or `packages/google/openiap/src/main/java/dev/hyo/openiap/Types.kt`
@@ -104,9 +106,10 @@ When fallback is required:
    acceptance criteria, changed-path conventions, and existing commit/push
    authority. Explicitly request one pass so `review-pr` remains the only polling
    owner.
-3. Do not let the fallback round re-enter `review-pr`, request reviewers, or
-   schedule its own five-minute loop. It may inspect current review/CI evidence,
-   but this workflow owns thread handling and polling.
+3. Do not let the fallback round re-enter `review-pr`, request reviewers, handle
+   trigger comments, invoke this fallback again, or schedule any recurring loop.
+   It may inspect current review/CI evidence, but this workflow owns thread
+   handling and polling.
 4. Fix and verify every validated finding using the normal response rules. If a
    fix changes the head, request CodeRabbit again after the fix batch and run
    fallback again only if it remains unavailable for the new head.

.claude/skills/review-self/SKILL.md

@@ -1,26 +1,26 @@
 ---
 name: review-self
-description: Independently review and improve the agent's current implementation, working-tree changes, or pull request; fix actionable in-scope gaps; rerun relevant verification; and recheck at five-minute intervals until the work is stable or genuinely blocked. Use when the user says "review-self", asks Claude to review its own changes, requests a self-review loop, or wants current work monitored for new issues after implementation.
+description: Independently review and simplify the agent's current implementation, working-tree changes, or pull request; enforce KISS and repository SSOT rules, fix actionable in-scope gaps, rerun relevant verification, and recheck at the user-requested interval (five minutes by default) until stable or genuinely blocked. Use when the user says "review-self", asks Claude to review its own changes, requests a self-review loop, or wants current work monitored for new issues after implementation.
 ---
 
 # Review Self (Claude Code)
 
 The canonical loop definition lives in `.codex/skills/review-self/SKILL.md`.
 Read it and follow every section — authority and scope preservation, target
-establishment, the review round, related OpenIAP workflows, the five-minute
-recheck contract, safe stopping conditions, and per-round communication are
-agent-agnostic and apply as written.
+establishment, the KISS/SSOT review round, related OpenIAP workflows, the
+requested-interval recheck contract, safe stopping conditions, and per-round
+communication are agent-agnostic and apply as written.
 
 ## Claude Code Notes
 
 - Where the canonical file routes through `$openiap-workflows`, read the
   matching `.claude/commands/*.md` file directly (or use the
   `.claude/skills/openiap-workflows` skill).
-- For the five-minute recheck, use Claude's real wake-up mechanism for the
-  current surface (for example a scheduled reminder / wake-up tool in Cowork
-  or the Agent SDK). If no such mechanism is available in the current session,
-  complete the current round and report that automatic re-entry could not be
-  scheduled — never emulate the loop with `sleep 300`, `while true`, or an
-  abandoned background process.
+- For interval rechecks, use Claude's real wake-up mechanism for the current
+  surface (for example a scheduled reminder / wake-up tool in Cowork or the
+  Agent SDK). Honor the user's interval, defaulting to five minutes. If no such
+  mechanism is available in the current session, complete the current round and
+  report that automatic re-entry could not be scheduled — never emulate the
+  loop with `sleep`, `while true`, or an abandoned background process.
 - Use read-only subagents (Task/Agent tool with an Explore-style agent) for
   independent review lenses on large or cross-cutting diffs.

.codex/skills/openiap-workflows/SKILL.md

@@ -69,6 +69,8 @@ appropriate labels before merging.
 
 ## Non-Negotiables
 
+- Apply the canonical KISS/SSOT release criteria in
+  `knowledge/internal/03-coding-style.md`.
 - Before any public GitHub write, apply the English-only communication guard in
   `knowledge/internal/06-git-deployment.md`. Private maintainer conversation
   language must never leak into issue, PR, review, release, or commit prose.

.codex/skills/review-self/SKILL.md

@@ -1,12 +1,13 @@
 ---
 name: review-self
-description: Independently review and improve Codex's current implementation, working-tree changes, or pull request; fix actionable in-scope gaps; rerun relevant verification; and recheck at five-minute intervals until the work is stable or genuinely blocked. Use when the user says "review-self", asks Codex to review its own changes, requests a self-review loop, or wants current work monitored for new issues after implementation.
+description: Independently review and simplify Codex's current implementation, working-tree changes, or pull request; enforce KISS and repository SSOT rules, fix actionable in-scope gaps, rerun relevant verification, and recheck at the user-requested interval (five minutes by default) until stable or genuinely blocked. Use when the user says "review-self", asks Codex to review its own changes, requests a self-review loop, or wants current work monitored for new issues after implementation.
 ---
 
 # Review Self
 
-Review the current work immediately, fix validated gaps, and use five-minute
-confirmation rounds until the result is stable.
+Review the current work immediately, fix validated gaps, and use confirmation
+rounds at the user's requested interval, or five minutes by default, until the
+result is stable.
 
 ## Preserve Authority And Scope
 
@@ -51,15 +52,18 @@ confirmation rounds until the result is stable.
    - public contracts, naming, compatibility, generated-file rules, and
      cross-package or SDK parity;
    - missing or weak tests, documentation, examples, migrations, and operational
-     safeguards required by the change.
+     safeguards required by the change;
+   - the canonical KISS/SSOT release rules in
+     `knowledge/internal/03-coding-style.md`.
 4. Use independent read-only subagents for separate review lenses when the diff
    is large or cross-cutting. Give them the raw target and request, not suspected
    findings or an expected answer.
 5. Validate every finding against the current code and applicable instructions.
    Reject pure taste, cosmetic churn, duplicate findings, and unrelated
    nice-to-have work.
-6. Fix all validated in-scope findings in one coherent batch. Regenerate generated
-   files only through their documented generator and preserve unrelated edits.
+6. Fix all validated in-scope findings in one coherent batch under those
+   canonical KISS/SSOT rules. Regenerate generated files only through their
+   documented generator and preserve unrelated edits.
 7. Re-read the resulting diff, then run the path-specific lint, typecheck, tests,
    builds, audits, and `git diff --check` required by the repository. Do not rerun
    an expensive unchanged check fingerprint unless new state can affect it.
@@ -105,7 +109,7 @@ the current PR head:
   requirements, and acceptance criteria. Preserve the commit/push and external
   write authority supplied by the calling workflow.
 - Do not re-enter `review-pr`, request external reviewers, handle its trigger
-  comments, invoke this fallback again, or schedule this skill's five-minute
+  comments, invoke this fallback again, or schedule this skill's recurring
   loop. `review-pr` remains the sole thread-handling and polling owner.
 - Return the reviewed head and working-tree fingerprints, validated findings and
   fixes, checks run, and a clean or blocked result. The caller may cache a clean
@@ -114,38 +118,42 @@ the current PR head:
 This single-round override prevents nested polling loops while still replacing
 the missing external review coverage with the full self-review procedure.
 
-## Recheck Every Five Minutes
+## Recheck At The Requested Interval
 
 - Run the first round immediately; never wait before the initial review.
 - If the user explicitly requests one pass, finish after that round and do not
   schedule a confirmation.
 - Use the product's real recurring-monitor or wake-up mechanism to schedule the
-  next round for 300 seconds after the current round. Make the scheduled prompt
-  explicitly invoke `$review-self`; re-enter this skill rather than trying to
-  perform semantic review inside a shell loop.
+  next round after the user's explicit interval, or 300 seconds by default.
+  Make the scheduled prompt explicitly invoke `$review-self`; re-enter this
+  skill rather than trying to perform semantic review inside a shell loop.
 - Keep at most one outstanding wake-up for the same target. Cancel or supersede
   stale duplicate wake-ups when the mechanism supports it.
 - Carry a compact state capsule in the scheduled prompt or monitor state. Include
   the original goal and acceptance criteria, target and base, head/tree and
   working-tree fingerprints, seen feedback and check IDs, poll count, clean
-  count, start time, finding fingerprints with fix attempts, and the existing
-  commit/push authority.
+  count, start time, the requested interval, the active interval, the current
+  pending-state fingerprint and first-seen time, finding fingerprints with fix
+  attempts, and the existing commit/push authority.
 - Keep loop state out of tracked repository files. Revalidate it against disk and
   remote state on every wake-up.
+- Validate the requested and active intervals on every wake-up. Preserve an
+  explicit user interval unless the user changes it; otherwise use the default.
 - Increment the clean count only after a complete round has no actionable
   findings, all requested verification passes, required checks are terminal and
   successful or explicitly allowed to skip, no actionable feedback remains, and
   the final diff has been reread. Reset it when any material state or finding
   changes, then allow the fully clean post-change result to start a new count.
-- Finish successfully after two consecutive clean snapshots separated by a
-  five-minute interval. Continue without a fixed round cap while new findings or
+- Finish successfully after two consecutive clean snapshots separated by the
+  active interval. Continue without a fixed round cap while new findings or
   state changes lead to meaningful progress.
-- Treat pending CI or review automation as neither clean nor failed. Poll it every
-  five minutes without repeating expensive local checks on an unchanged
-  fingerprint.
+- Treat pending CI or review automation as neither clean nor failed. Poll it at
+  the active interval without repeating expensive local checks on an unchanged
+  fingerprint. Reset its first-seen time whenever the pending-state fingerprint
+  changes.
 - If no recurring mechanism is available, complete the current round and report
   that automatic re-entry could not be scheduled. Do not emulate it with
-  `sleep 300`, `while true`, `nohup`, or an abandoned background process, and
+  `sleep`, `while true`, `nohup`, or an abandoned background process, and
   never claim a future pass was scheduled unless the mechanism actually accepted
   it.
 
@@ -161,7 +169,7 @@ Stop the loop and report the exact state when any of these conditions holds:
 - the same root finding remains after two fix attempts;
 - the same authentication, rate-limit, tool, or environment failure blocks three
   consecutive rounds;
-- only unchanged pending external state remains after 12 polls (about one hour).
+- only unchanged pending external state remains for at least one hour.
 
 Do not call a blocked or interrupted result clean. Do not stop merely because one
 poll is a no-op while asynchronous state is still pending.

.codex/skills/review-self/agents/openai.yaml

@@ -1,4 +1,4 @@
 interface:
   display_name: "Review Self"
-  short_description: "Review current work in five-minute loops"
-  default_prompt: "Use $review-self to review the current work, fix actionable findings, and recheck it every five minutes until stable."
+  short_description: "Review and simplify current work until stable"
+  default_prompt: "Use $review-self to review and simplify the current work under KISS and SSOT rules, fix actionable findings, and recheck at the requested interval until stable."

AGENTS.md

@@ -74,6 +74,12 @@ Before posting or editing issues, pull requests, reviews, discussions, commits,
 release notes, or GitHub Releases, follow the mandatory language guard in
 [`knowledge/internal/06-git-deployment.md`](knowledge/internal/06-git-deployment.md#public-github-communication-language).
 
+### KISS and SSOT
+
+KISS and SSOT are mandatory release criteria. The canonical rules live in
+[`knowledge/internal/03-coding-style.md`](knowledge/internal/03-coding-style.md#0-kiss-and-ssot-are-release-requirements).
+Apply that section before implementation and during every review.
+
 ### Platform Function Naming
 
 - **iOS functions**: Must end with `IOS` suffix (e.g., `syncIOS`, `getReceiptDataIOS`)
@@ -255,7 +261,8 @@ keep its wording agent-neutral.
 1. Reviews the complete current diff, including staged, unstaged, and untracked
    work, against the original request and repository conventions
 2. Fixes validated in-scope findings and runs path-specific verification
-3. Rechecks through a real recurring wake-up after five minutes
+3. Rechecks through a real recurring wake-up at the user's requested interval,
+   defaulting to five minutes
 4. Finishes after two consecutive clean snapshots, or reports the exact blocker
 
 `review-self` does not grant commit, push, PR, merge, deploy, or release authority

knowledge/_claude-context/context.md

@@ -1,7 +1,7 @@
 # OpenIAP Project Context
 
 > **Auto-generated for Claude Code**
-> Last updated: 2026-08-09T13:54:08.141Z
+> Last updated: 2026-08-09T21:46:58.647Z
 >
 > Usage: `claude --context knowledge/_claude-context/context.md`
 
@@ -541,6 +541,28 @@ isBillingProgramAvailableAndroid(program: BillingProgramAndroid!): BillingProgra
 
 ## General Principles
 
+### 0. KISS and SSOT Are Release Requirements
+
+Prefer the simplest correct design that satisfies verified requirements. KISS
+does not justify skipping error handling, lifecycle safety, tests, or public
+contracts; it requires meeting them with the fewest independent concepts.
+
+- Give each stateful resource one clear owner and one terminal cleanup path.
+  Pass or reference that owner instead of creating fallback stores, scopes,
+  caches, or managers at multiple layers.
+- Keep each fact in one canonical source. Generate or link mirrors and adapters;
+  never maintain equivalent rules, versions, schemas, or lifecycle decisions in
+  parallel files.
+- Reuse an existing abstraction when it already owns the invariant. Add a new
+  helper only for real reuse, a necessary platform boundary, or isolated testing;
+  keep single-use helpers local.
+- Do not add speculative configuration, indirection, background work, or state.
+  Every new layer must name the invariant it protects, its owner, its cleanup,
+  and the test that proves it is needed.
+- When fixing a bug, first look for state or code that can be deleted or
+  consolidated. Prefer one understandable path over several defensive fallback
+  paths.
+
 ### 1. Explicit Over Implicit
 
 Always be explicit about types and intentions:

knowledge/internal/03-coding-style.md

@@ -5,6 +5,28 @@
 
 ## General Principles
 
+### 0. KISS and SSOT Are Release Requirements
+
+Prefer the simplest correct design that satisfies verified requirements. KISS
+does not justify skipping error handling, lifecycle safety, tests, or public
+contracts; it requires meeting them with the fewest independent concepts.
+
+- Give each stateful resource one clear owner and one terminal cleanup path.
+  Pass or reference that owner instead of creating fallback stores, scopes,
+  caches, or managers at multiple layers.
+- Keep each fact in one canonical source. Generate or link mirrors and adapters;
+  never maintain equivalent rules, versions, schemas, or lifecycle decisions in
+  parallel files.
+- Reuse an existing abstraction when it already owns the invariant. Add a new
+  helper only for real reuse, a necessary platform boundary, or isolated testing;
+  keep single-use helpers local.
+- Do not add speculative configuration, indirection, background work, or state.
+  Every new layer must name the invariant it protects, its owner, its cleanup,
+  and the test that proves it is needed.
+- When fixing a bug, first look for state or code that can be deleted or
+  consolidated. Prefer one understandable path over several defensive fallback
+  paths.
+
 ### 1. Explicit Over Implicit
 
 Always be explicit about types and intentions:

libraries/expo-iap/src/types.ts

@@ -46,7 +46,12 @@ export interface AdvancedCommerceInfoIOS {
   estimatedTax?: (string | null);
   /** The items purchased as part of this transaction */
   items: AdvancedCommerceItemIOS[];
-  /** Subscription period for this transaction */
+  /**
+   * Subscription period for this transaction.
+   * Available in OpenIAP Spec 3.1.0 / openiap-apple 3.1.0
+   * (requires iOS 18.4+, macOS 15.4+, tvOS 18.4+, watchOS 11.4+,
+   * or visionOS 2.4+).
+   */
   period?: (SubscriptionPeriodValueIOS | null);
   /** Request reference identifier for tracking */
   requestReferenceId?: (string | null);

libraries/flutter_inapp_purchase/lib/types.dart

@@ -1238,7 +1238,10 @@ class AdvancedCommerceInfoIOS {
   final String? estimatedTax;
   /// The items purchased as part of this transaction
   final List<AdvancedCommerceItemIOS> items;
-  /// Subscription period for this transaction
+  /// Subscription period for this transaction.
+  /// Available in OpenIAP Spec 3.1.0 / openiap-apple 3.1.0
+  /// (requires iOS 18.4+, macOS 15.4+, tvOS 18.4+, watchOS 11.4+,
+  /// or visionOS 2.4+).
   final SubscriptionPeriodValueIOS? period;
   /// Request reference identifier for tracking
   final String? requestReferenceId;

libraries/godot-iap/addons/godot-iap/types.gd

@@ -439,7 +439,7 @@ class ActiveSubscription:
 class AdvancedCommerceInfoIOS:
 	## The items purchased as part of this transaction
 	var items: Array[AdvancedCommerceItemIOS] = []
-	## Subscription period for this transaction
+	## Subscription period for this transaction. Available in OpenIAP Spec 3.1.0 / openiap-apple 3.1.0 (requires iOS 18.4+, macOS 15.4+, tvOS 18.4+, watchOS 11.4+, or visionOS 2.4+).
 	var period: SubscriptionPeriodValueIOS
 	## Request reference identifier for tracking
 	var request_reference_id: Variant = null

libraries/kmp-iap/library/src/commonMain/kotlin/io/github/hyochan/kmpiap/openiap/Types.kt

@@ -1387,7 +1387,10 @@ public data class AdvancedCommerceInfoIOS(
      */
     val items: List<AdvancedCommerceItemIOS>,
     /**
-     * Subscription period for this transaction
+     * Subscription period for this transaction.
+     * Available in OpenIAP Spec 3.1.0 / openiap-apple 3.1.0
+     * (requires iOS 18.4+, macOS 15.4+, tvOS 18.4+, watchOS 11.4+,
+     * or visionOS 2.4+).
      */
     val period: SubscriptionPeriodValueIOS? = null,
     /**