Repository navigation
fix(chat): say the tool-risk badge's vote is feedback only - #32077
Conversation
df8056f to
71ab58d
Compare
|
Rebased: main moved in docs/system-specs/modules/decisions.md, website/src/i18n/en.context.json and website/src/i18n/locales/*.json since this head went green (76ec06a -> a3fb5af); CI re-running. The patch is unchanged ( [operator: bolichen97] |
Design Review (Opus 5.5) — ✅ PASSDesign-level review of Design-Verdict: PASS Small copy and layout fix with rendered evidence; the renamed key has no readers outside the patch. [DESIGN-REVIEWED] eb2958a Description read[DESCRIPTION-READ] 49b7fec9af2c95e8189e0f1fa68c4e89c658933779080c4d3b0c0fcba56978b4 That sha256 names everything this verdict read: the PR description, plus the 1 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with |
First Principles Review (Opus 5.5) — ✅ PASSPremise-level review of First-Principles-Verdict: PASS Nothing blocks: a copy fix tied to issue #12298, backed by a test that fails on base, with the key rename required by the catalog rule. What this change shipsInventory (4 items) — 4 justifiedIntent: make it visible on the tool-risk badge that a thumbs vote is only feedback and does not allow or block the call. This is a FIX.
[FIRST-PRINCIPLES-REVIEWED] eb2958a Description read[DESCRIPTION-READ] ad2e7f19b0776c8234013a43adde7d8af84e87e685d481e744f0ff5d273cab29 That sha256 names the PR description this verdict read. Recompute it with |
Opus 5.5 Review — ✅ no blocking findingsReviewed Verdict parsed from the review's SHA-scoped output markers for commit False positive or not applicable? A repository writer can comment: |
UX Review (Opus 5.5) — ✅ PASSUX-level review of UX-Verdict: PASS Nothing blocks: the screenshots show the new label and the wrapped 320 px layout, and the blind reader read the thumbs correctly as a feedback-only vote. Notes
[UX-REVIEWED] eb2958a Description read[DESCRIPTION-READ] 1aab04de48738510a621bff6263f247114b05572cdfdba3f1db9c49e00e2e158 That sha256 names everything this verdict read: the PR description, plus the 8 evidence file(s) the workflow collected from it and handed this review. It covers that evidence because the media strip replaces every attachment URL with the same placeholder, so an edit swapping one attachment for another would leave the description's captured text identical while what this review judged changed. Recompute it with |
GPT 6.1 Review — ✅ no blocking findingsGPT 6.1 completed its review of This comment is updated in place on each push. Review detailsNo findings. False positive or not applicable? A repository writer can comment: |
71ab58d to
b6605e3
Compare
|
Rebased: main moved in website/src/i18n/en.context.json, website/src/i18n/locales/*.json and docs/system-specs/modules/ since this head went green (a3fb5af -> 4d0ae46); CI re-running. The patch is unchanged (same [operator: bolichen97] |
The risk badge's thumbs carried the label "Jev", repeating the word the line opens with, and nothing visible said a vote is only feedback: that meaning lived in the buttons' title and aria-label, which a sighted reader on touch or keyboard never sees. The label now reads "Feedback only". The key is renamed from rate_jev to rate_label because its meaning changed, and all twelve catalogs, the pseudolocale and the translator context carry the new value. The badge row wraps, so on a phone-width pane the note and the thumbs take their own lines instead of the note being squeezed to a narrow column. The decisions spec states the label. Closes #12298
b6605e3 to
eb2958a
Compare
|
Rebased: main moved in website/src/i18n/locales/*.json, docs/system-specs/modules/ and test/test_chat_utils.py since this head was measured (4d0ae46 -> bab2387); CI re-running. The red Backend Tests shard 2 (and the Coverage Gate that follows it) on the old head was main's own [operator: bolichen97] |
KiroCrewCaptain
left a comment
There was a problem hiding this comment.
Tier 1 auto-approve: fix (17 files). Criteria: no conflict, no requested changes, security path denylist clean, design-doc gate clean, security checklist all-NO, no reviewer-lane findings, readiness signal fresh for this head SHA, declared category matches the diff. Category: replaces the tool-risk badge's rating label with a 'Feedback only' string in every locale; with an updated test. Spec files changed as a ride-along (a minority of the diff on both file count and changed lines), not reviewed as a design decision: docs/system-specs/modules/decisions.md.
The risk badge's thumbs carried the label "Jev", repeating the word the line opens with, and nothing visible said a vote is only feedback: that meaning lived in the buttons' title and aria-label, which a sighted reader on touch or keyboard never sees. The label now reads "Feedback only". The key is renamed from rate_jev to rate_label because its meaning changed, and all twelve catalogs, the pseudolocale and the translator context carry the new value. The badge row wraps, so on a phone-width pane the note and the thumbs take their own lines instead of the note being squeezed to a narrow column. The decisions spec states the label. Closes #12298 (cherry picked from commit 84be3f2)
Problem / Motivation
Goal: The tool-risk badge's thumbs say in visible text that a vote is only feedback, without repeating Jev's name.
The badge reads
Jev: risky (0.88) flag only, did not allow or block Jev 👍 👎. The second "Jev" repeats the word the line opens with. Nothing visible says a vote is only feedback: that lives in the buttons'titleandaria-label, which a sighted reader on touch or keyboard never sees. These are the last two open residuals of #12298. The rest of the issue already shipped: item 1's wording in #16320, item 2 documented as a v1 limit in the feature map, the visible flag-only note in #23869, and the Decisions-card capture harness now stubsdecisions_enabled.Why it matters
A thumbs pair right after "risky" on a tool card can read as an allow or block control. The badge exists to say the flag changed nothing, so its one control must not suggest the opposite.
Not a goal
DecisionVerdictThumbscontrol.What changed (motivation → approach → change)
The thumbs label now reads Feedback only instead of a second "Jev". It is short on purpose: a longer "Rate this flag (feedback only)" pushed the thumbs past the edge of a 320 px screen in Italian, German, French and Bengali.
pages/chat/ToolRiskBadge.tsxflex-wrap, so on a phone the note and the thumbs take their own lines. Without it the note is squeezed to a narrow column: at 320 px the caution badge's note was 27 px wide and seven lines tall, and is now 180 pxi18n/locales/*.json,en.context.jsontoolRiskBadge.rate_jevrenamed totoolRiskBadge.rate_labelbecause its meaning changed (catalog rule "One key, one meaning"); all twelve languages translated in each style guide's register, pseudolocale regenerated, translator context rewrittendocs/system-specs/modules/decisions.mdBackwards compatibility
Compatible: a display string and its catalog key change; no stored data, route or config is involved.
Reader: website/src/pages/chat/ToolRiskBadge.tsx:ToolRiskBadge -- chat -- ToolRiskBadge.test.tsx (the only reader of the renamed key;
[key-refs]reports 0 dangling)Tests
website/src/test/ToolRiskBadge.test.tsx: the visible label is "Feedback only" and the badge's text says "Jev" exactly once. It fails on main (expected 'Jev' to be ...) and passes here, five runs out of five with--sequence.shuffle(seeds 11 to 55).npm run i18n:checkwith the base ref):[changed-values]and[changed-passthrough]0,[key-refs]and[pseudolocale]clean; the twelve locale style tests pass.Proof: not provable -- frontend test (prove.py runs Python only)
Manual verification
Built the SPA and ran
scripts/capture-tool-risk-badge.mjsagainst it at 1280 px (one line, as before), then the same scene at 320 px in all twelve languages: the badge'sscrollWidthequals itsclientWidthin every one, and the thumbs end inside the badge.Screenshots / video
Risky badge, light, and caution badge, dark, at 1280 px:
Caution badge at 320 px, English and Simplified Chinese:
Whole transcript, light
Related Issues
Closes #12298
Sibling sweep
Swept: git grep -n "label={i18nT(" -- website/src/pages/chat
Sibling: website/src/pages/chat/CompactionKeepLine.tsx -- not affected -- its label already names the act, not Jev
Sibling: website/src/pages/chat/SteerDecisionLine.tsx -- not affected -- its label already names the choice, not Jev
Sibling: website/src/pages/chat/DecisionStrip.tsx -- not affected -- two pairs share one strip there, so the Jev name tells them apart
Pattern harvest
Not generalizable: a copy residual on one surface; the shared control already requires a visible label, and choosing its words is per-surface judgement.
Checklist
feat|fix|docs|style|refactor|perf|test|chore|ci|build|revert: ...)