[e54-bot]

Fixes #142

What

One test, crates/athena-sim/tests/enum_reads.rs. It lists every catalog value entry whose return type is an enum domain, then runs Compare(<read>, ==, <enum literal>) for each modeled read and asserts true.

How the read set is derived

athena-sim has no registry of modeled reads: dispatch is match name arms in call.rs and player.rs. So the set comes from the workshop-rs catalog (entries_of(Kind::Value) with return_type() in an enum domain; the Vector domain is excluded because its reads yield numeric vectors).

A new catalog read in that set fails the test until it is added to MODELED or UNMODELED. A new enum-valued read therefore cannot go uncovered, but it needs a one-line table entry, because the call arguments cannot be generated. A read in UNMODELED must not complete, so modeling it later fails the test and forces a move to MODELED.

Reads covered

Modeled, compared with the literal: hero, getHero (Hero Of), teamOf, currentMap, getCurrentGamemode, gameMode, workshop setting hero (createWorkshopSettingHero). The scenario uses ana, team1, king's row, skirmish, none equal to the member ids.
Unmodeled, asserted not to complete: oppositeTeamOf, getHeroOfDuplication, getControlScoringTeam, eventAbility, customColor.

Bug found

None. Every modeled read is true on current main.

Mutation evidence

Changed "getHero" => player.hero_member.clone() to player.hero.clone() in player.rs (authored spelling), ran the test, then git checkout:

every_enum_valued_read_equals_its_enum_literal --- FAILED
assertion `left == right` failed: getHero: Hero Of(Host Player)
  left: Bool(false)
 right: Bool(true)

Verification

• cargo test: all pass
• cargo clippy --all-targets -- -D warnings: clean
• cargo fmt --check: clean

[e54-bot]

Verdict: LGTM (head 57cb085)

Independent check, not from the PR's own tests.

1. Derivation. I listed entries_of(Kind::Value) with an enum-domain return type from the catalog. Excluding Vector leaves 12 reads: hero, getHero, teamOf, oppositeTeamOf, customColor, currentMap, eventAbility, getHeroOfDuplication, createWorkshopSettingHero, getControlScoringTeam, getCurrentGamemode, gameMode. MODELED (7) + UNMODELED (5) equal this set exactly. Only Vector is a non-member-id enum domain among the catalog's enum-returning values, so no other exclusion is needed. Deleting the eventAbility table line fails with "enum-valued read eventAbility is in neither MODELED nor UNMODELED".
2. Classification. All 5 UNMODELED reads stop with NotModeled for the read itself (value oppositeTeamOf, value getHeroOfDuplication, value getControlScoringTeam, value eventAbility), not for a missing scenario key. customColor is modeled as a Display token (call.rs, athena#113) and the Compare against Color(Red) is rejected as NotModeled. That is sound: a display token has no member id. Every read that dispatch models and that returns an enum is in MODELED.
3. Spellings. The scenario uses ana, team1, king's row, skirmish; the member ids differ. getHero, teamOf, currentMap and getCurrentGamemode exercise the member-id path. hero, gameMode and createWorkshopSettingHero are literal constructors, so the scenario spelling does not reach them. This is cheap, harmless coverage; no change needed.
4. Mutation. getHero => player.hero.clone() fails (getHero: Hero Of(Host Player), left Bool(false), right Bool(true)). Team Of returning a lowercased debug spelling fails the same way (teamOf: Team Of(Host Player)). Both restored.
5. Size and shape. One file, no new dependency, named for behavior. The existing Hero Of tests check specific reads, so the table-driven test does not duplicate them.
6. Gates. cargo test passes. cargo clippy --all-targets -- -D warnings is clean. cargo fmt --check is clean. git diff --check origin/main...HEAD is clean. CI passes (Quality, WebAssembly build). The workflow lint job is skipped. mergeStateStatus is CLEAN.