Fix contradiction hunt sensitivity disclosure

[bob-codedaptive]

Motivation

• The contradiction-hunt pass returned raw borderline snippets from hydrated drawers without applying a recall sensitivity/exportability ceiling, exposing up to 160 characters of restricted/secret memory via the MCP tools.
• The goal is to restore the default recall disclosure posture so MCP-exposed hunt results cannot leak high-sensitivity content.

Description

• Swift: added a sensitivity gate in GeniusLocusKit.huntContradictions to skip candidate drawers whose adjectiveSensitivity is above .elevated and added isHuntRecallEligible helper to centralize the ceiling check (packages/kits/GeniusLocusKit/Sources/GeniusLocusKit/Brain/ContradictionHunt.swift).
• Rust: applied the same gate in the Rust mirror (hunt_contradictions) by skipping drawers whose adjective_sensitivity() exceeds the Elevated tier (packages/kits/GeniusLocusKit/rust/src/coordinator.rs).
• Tests: added parity regression tests in Swift and Rust that plant restricted/secret drawers and assert the hunt returns no proposed or borderline snippets (Swift test added to ContradictionHuntTests.swift; Rust test added to coordinator.rs).

Testing

• Attempted swift test --filter ContradictionHuntTests but execution was blocked by SwiftPM dependency fetch failures (HTTP 403) so tests could not complete in this environment.
• Attempted cargo test for the new Rust test but execution was blocked by crates.io index fetch failures (HTTP 403) so tests could not complete in this environment.
• Ran cargo fmt --check; pre-existing repository formatting drift produced failures unrelated to this change and prevented a clean check in this environment.

---

Codex Task

[github-actions]

Thank you for your contribution to MOOTx01. Before it can be merged we need you to sign our Contributor License Agreement. You can read it here: CLA.md. To sign, post a comment on this pull request with exactly the following text:

---

I have read the CLA Document and I hereby sign the CLA

---

_{You can retrigger this bot by commenting recheck in this Pull Request. Posted by the CLA Assistant Lite bot.}

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 26c07098fa

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Compare the raw sensitivity field, not the decoded enum

For drawers whose adjective bitmap contains a reserved/intermediate sensitivity raw value above elevated (for example a future tier at raw 24), drawer.adjectiveSensitivity falls back to .normal, so this check lets the pair reach ConflictCue and can still emit borderline snippets. Default recall's .sensitivityAtMost(.elevated) is enforced as a raw bitmap threshold, so this helper is not actually equivalent to the recall ceiling it is trying to mirror; compare the extracted raw field to 16 instead of comparing the lossy decoded enum.

Useful? React with 👍 / 👎.

[chatgpt-codex-connector]

Compare the raw sensitivity field, not the decoded enum

When a drawer has a reserved/intermediate adjective-sensitivity raw value above Elevated, adjective_sensitivity() decodes it through from_raw and falls back to Normal, causing this gate to allow the pair even though the default recall ceiling uses a raw SensitivityAtMost(Elevated) threshold and would exclude it. This leaves a disclosure gap for newer/imported rows using reserved sensitivity values; the hunt gate should compare the raw bits from adjective_bitmap directly.

Useful? React with 👍 / 👎.

[bob-codedaptive]

Closing as a duplicate of #20, now merged. Both applied the same sensitivity-ceiling guard at the contradiction-hunt disclosure chokepoint in both legs; #20 was kept as it preserved the endpoint-stamping comment. (#19 already closed.)