feat(serve): variant scoping for 8 handlers (close incoherence flagged in PR #215)

[avrabe]

Summary

Closes the variant-scoping incoherence flagged by PR #215's coverage audit and pinned by tests/playwright/rendering-invariants.spec.ts test #8: 8 dashboard handlers silently dropped ?variant= while the layout banner showed scoped, leaving a surprising asymmetry between what the URL claimed and what the handler actually rendered.

This is exactly the kind of "user signal is consistency" fix the audit flagged: every other ViewParams-bearing route (/artifacts, /coverage, /matrix, /stats, /stpa, /validate) was already scoped — these eight just lagged behind.

Per-handler design choices

Bias toward A: accept-and-scope since variant scoping is semantically meaningful nearly everywhere, and the renderer-side machinery (VariantScope::render_context) already exists and is identical to what the seven scoped routes use.

Handler	Choice	Rationale
graph_view	A	Variant scoping reduces the node count, which actually helps the O(n^2)-ish layout pass on large dogfood projects (where the existing comment notes ~57s renders). Performance argument, not just consistency.
verification_view	A	Verification status reflects the active variant's artifact slice — minimal-ci should not show verification rows for artifacts not in scope.
eu_ai_act_view	A	Annex IV compliance is naturally per-variant (different variants ship different artifact mixes — full-desktop covers more sections than minimal-ci).
traceability_view	A	The canonical scope-sensitive view; was the most surprising silent-drop.
doc_linkage_view	A	Per-doc link counts now derive from the scoped store.
documents_list	A	Document list itself isn't strictly per-variant, but the link counts shown for each doc are scoped.
results_view	A	Run results are tied to artifacts; scoping to the variant's bound IDs is meaningful.
search_view	A	Search now ranks only against in-scope artifacts. (/search is fragment-only so no banner; the param is honored upstream of layout.)
externals_list	C (explicit ignore + comment)	External projects are loaded from sibling repos and don't participate in this project's feature-model bindings, so variant filtering has no semantic meaning there. Param is accepted gracefully (so bookmarked URLs degrade) and the rationale is documented inline.

What changed

rivet-cli/src/serve/views.rs — the eight handlers above. Each one now:

• Takes ViewParams (or extends its existing param struct with variant: Option<String>).
• Calls try_build_scope(&state, &params.variant) — the same helper used by artifacts_list, coverage_view, matrix_view, etc.
• Returns Response (not Html<String>) so the 400-on-unknown-variant error path matches the rest of the dashboard.
• For externals_list, accepts ViewParams but ignores the variant field — documented in the doc-comment.

tests/playwright/rendering-invariants.spec.ts — flips test #8 from pinning the silent-drop bug to asserting the new scoped behaviour, then adds two more end-to-end tests for /traceability and /documents.

rivet-cli/tests/serve_integration.rs — three new Rust integration tests:

• newly_scoped_handlers_render_variant_banner — sweeps the seven banner-routes, asserts 200+banner on valid variant and 400 on unknown variant.
• search_handler_honors_variant_scope — same contract for /search minus the banner (fragment-only endpoint).
• externals_view_accepts_variant_param_without_400 — pins the Choice C contract for /externals.

Test plan

• cargo build -p rivet-cli — clean
• cargo test -p rivet-cli — 99 lib + 42 serve_integration + the rest pass
• cargo clippy -p rivet-cli --all-targets -- -D warnings — clean (only pre-existing MSRV note)
• cargo fmt -p rivet-cli -- --check — clean
• CI: tests/playwright/rendering-invariants.spec.ts test feat: cross-repository artifact linking #8 plus two new tests
• CI: end-to-end Playwright suite

Trailers

Refs: FEAT-001 (dashboard/serve) per CLAUDE.md quick reference.

[github-actions]

⚠️ Performance Alert ⚠️

Possible performance regression was detected for benchmark 'Rivet Criterion Benchmarks'.
Benchmark result of this commit is worse than the previous benchmark result exceeding threshold 1.20.

Benchmark suite	Current: af1c7a7	Previous: 3cdb942	Ratio
query/10000	154546 ns/iter (± 2066)	110920 ns/iter (± 985)	1.39

This comment was automatically generated by workflow using github-action-benchmark.

[github-actions]

📐 Rivet artifact delta

No artifact changes in this PR. Code-only changes (renderer, CLI wiring, tests) don't touch the artifact graph.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!