fix(lint): stop scene-exit hard-kill rules from contradicting gsap_animates_clip_element

[miguel-heygen]

Root-caused from two independent post-release feedback reports of the same lint contradiction.

The contradiction

scene_layer_missing_visibility_kill and gsap_exit_missing_hard_kill tell you to add tl.set(selector, { visibility: "hidden" }, t) on an exiting scene element. When that element is also class="clip", the exact tl.set they recommend is then flagged by gsap_animates_clip_element ("the framework manages clip visibility via visibility — do not animate these properties on clip elements"). Two users hit this independently:

"the linter both requires visibility:hidden on clip elements and forbids GSAP from animating visibility on clip elements."

"'scene_layer_missing_visibility_kill' lint rule tells you to add tl.set(...) on the exiting scene layer, but if that layer is also class='clip', the 'gsap_animates_clip_element' rule then errors on the same visibility set. [...] Would help if the first rule's fix hint mentioned the inner-wrapper pattern for clip scenes."

That second report describes exactly the workaround this fix teaches: wrap the scene's content in an inner non-clip <div> and move the exit tween + hard kill onto that wrapper.

Fix

Both rules now detect when the flagged/exiting selector is a clip element and, only in that case, point the fixHint at the inner-wrapper pattern instead of a tl.set on the clip element itself:

• gsap_exit_missing_hard_kill reuses the clipIds/clipClasses maps already built in its enclosing rule.
• scene_layer_missing_visibility_kill (a separate, older id-pattern rule) checks the flagged tag's class list directly, since it doesn't share that closure.

Non-clip targets are unaffected — identical fix hint as before.

Tests

Two new cases in gsap.test.ts (one per rule) asserting the fix hint mentions the inner-wrapper pattern and does NOT suggest a tl.set on the clip selector. All 67 existing + new tests in the suite pass.

[james-russo-rames-d-jusso]

Reviewed at d9aacd3092 (batch review, Group C lint quality; layering on Magi's own vetting).
Note: bot-authored (Magi via miguel-heygen); COMMENT-quality — stamp routing per protocol.

Summary — reroutes the fixHint copy of gsap_exit_missing_hard_kill and scene_layer_missing_visibility_kill to the inner-wrapper pattern when the exiting/flagged selector is itself a clip element. Resolves the contradiction against gsap_animates_clip_element (which errors on tl.set(clip, { visibility })).

Verified

• Detection unchanged, only hint copy changes. The hasHardKill check and the finding emission run before the branch; if the author DID add a hard kill, both rules still exit early. Non-clip targets get the original hint verbatim. No false negatives introduced.
• Symmetric fix across two structurally-different rules. gsap_exit_missing_hard_kill reuses the clipIds / clipClasses maps already built in its enclosing closure (packages/lint/src/rules/gsap.ts:521-538). scene_layer_missing_visibility_kill is a separate top-level rule with no closure access, so it reads class attr on the flagged tag directly. Both routes converge on class="clip" — the same signal gsap_animates_clip_element uses at :678-680. Consistent.
• Scene-inside-clip nesting: <div class="clip"><div id="scene1"...> — scene1 isn't itself clip, isClip=false, gets the plain hint tl.set("#scene1", ...). Since scene1 isn't a clip element, gsap_animates_clip_element won't fire on it. No contradiction. Correct.
• Two tests, one per rule, assert (a) hint mentions "clip element" and "inner", (b) hint does NOT contain the direct tl.set on the clip selector. Targeted at both parts of the contradiction.

Nits

• 🟡 The inner-wrapper hint template still emits the numeric boundary time (boundary.toFixed(2)) even though the actual selector is a placeholder ("<inner-selector>"). Author needs to substitute two things; the timing is still a useful reference number. Fine as-is.
• 🟡 The hint text is duplicated between the two rules ("is a clip element — the framework already manages its visibility. Wrap the scene's content in an inner non-clip <div>..."). If a third rule joins this contradiction someday, worth pulling into a shared helper. Not for this PR.

Questions

• ↩️ Is there any consumer (a build step, docs generator, telemetry) that dedupes findings by exact fixHint string? If yes, the divergence between clip-case and non-clip-case hints for the same code would produce two hint-buckets in whatever dashboard reads them. Probably not — but worth a git grep 'fixHint' if you have downstream aggregation.

— Rames D Jusso

[jrusso1020]

APPROVE — verified CI green (0 fail / 0 pending) + no open CR at this head. Non-author stamp clearing the review gate on the Magi self-initiated draft-pass batch, which James greenlit and RDJ batch-cleared: both security holds re-verified at R2 (#1866 chrome-shell reclaim-race closed via the reclaim-gate + mtime recheck; #1845 Windows npx shell-injection closed — no cmd.exe, node <npx-cli.js> with pure argv), zero drift on the other nine, all green.

Merge via Magi's normal path (no admin-merge). Ordering note: the skills-manifest triple-conflict on #1877 / #1862 / #1845 (all bump hyperframes-media.hash) needs sequential rebase + regen at merge time; the rest land in any order.

— Rames Jusso