Repository navigation
fix: name the dashboard control for the refused app verb - #36345
Conversation
When a running gateway refused `kirocrew app update` or `kirocrew app uninstall`, the CLI told the operator to toggle the app in the dashboard, a remedy that fits only enable and disable. Each gateway verb now names its own dashboard control: the toggle for enable and disable, Update (or Sync) on the app's page for update, and Uninstall there for uninstall. Part of #26967
ab62747 to
144e223
Compare
|
Rebased onto
|
UX Review (Opus 5.5) — ⏭️ skippedRevision |
First Principles Review (Opus 5.5) — ✅ PASSPremise-level review of First-Principles-Verdict: PASS Nothing blocks. The table has four keys, and each key matches one of the four verbs that call the shared helper. What this change shipsInventory (3 items) — 3 justifiedIntent: when the gateway refuses
[FIRST-PRINCIPLES-REVIEWED] 144e223 Description read[DESCRIPTION-READ] ac16ecb17a21005dc7ad6e12cd2c4c279bb7ab324aedb6f0e631f6fde3bf7da5 That sha256 names the PR description this verdict read. Recompute it with |
Design Review (Opus 5.5) — ✅ PASSDesign-level review of Design-Verdict: PASS Nothing blocks: the fix sits at the one shared call site, and the gateway is only called with the four verbs the table covers. [DESIGN-REVIEWED] 144e223 Description read[DESCRIPTION-READ] 95aaee148e94ac75c30b6710585b69de127e804830eb10404493a13a5185c542 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 |
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: |
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: |
|
Head
|
KiroCrewCaptain
left a comment
There was a problem hiding this comment.
Tier 1 auto-approve: fix (5 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: a refused app lifecycle verb names the dashboard control for that verb, with tests. 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/cli.md.
Problem / Motivation
Goal: When a running gateway refuses
kirocrew app updateorkirocrew app uninstall, the CLI names the dashboard control for that verb instead of telling the operator to toggle the app.Adoption smoke 3 for #26967: with a gateway running,
kirocrew app update not-an-appprinted❌ gateway refused: app 'not-an-app' not installed. Toggle the app in the dashboard instead.Toggling an app does not update it. The same hint followed every refuseduninstall.Why it matters
The built-in app campaign moves apps out of core, so
kirocrew app updateis now a common way to apply an app's fix release. An operator whose update is refused is sent to a control that cannot do what they asked, and the dashboard's Update (or Sync) button, which carries the trust-the-repository consent the CLI has no flag for, goes unmentioned.Not a goal
installhint, which already has its own wording.What changed (motivation → approach → change)
Symptom: a refused
app updateorapp uninstalltold the operator to toggle the app. Cause:_run_app_action_through_gateway, shared by enable, disable, update and uninstall, passed one fixed hint. Change: the hint is looked up per verb in_DASHBOARD_REMEDY_BY_ACTION. Enable and disable keep the toggle wording. Update names Update (or Sync) on the app's page, the wording the native-Windows refusal of the same verb already uses. Uninstall names Uninstall on that page. Theclispec row forkirocrew appnow states this, andsrc/kiro_crew/cli_commands.pylists the new name in its type-checking imports of the facade's served names.Backwards compatibility
Compatible: only the hint after a gateway refusal changes for
updateanduninstall. Exit codes, the refusal reason and the enable and disable text are unchanged.Removes nothing: no path, key, record or function is taken away; one stderr sentence is reworded for two verbs.
Tests
test_cli_app_update.py::TestCliUpdate::test_a_refusal_names_the_update_control_not_the_toggle: a refused update names Update (or Sync) and never says Toggle.test_apps_uninstall_backend_stop.py::TestCliUninstallDelegatesToTheGateway::test_a_refusal_names_the_uninstall_control_not_the_toggle: a refused uninstall names Uninstall, never says Toggle, and runs no file-only uninstall.test_gateway_mint_error_exits_with_dashboard_hintstill pins the toggle wording fordisable.Proof: PROVEN test/test_apps_uninstall_backend_stop.py::TestCliUninstallDelegatesToTheGateway::test_a_refusal_names_the_uninstall_control_not_the_toggle, test/test_cli_app_update.py::TestCliUpdate::test_a_refusal_names_the_update_control_not_the_toggle (head 144e223)
Determinism: both new tests passed 10 repeats in a row, and passed again run in reverse order ahead of
test_cli_commands_coverage.py. They fake the gateway call and read only captured stderr: no clock, port, sleep or ordering.Manual verification
N/A: unit coverage sufficient. The change is one stderr string per verb, and the tests drive the real
_handle_apppath down to the faked gateway call.Related Issues
Part of #26967
Sibling sweep
Swept: git grep -n "dashboard instead|hint=" -- src/kiro_crew/cli_commands*.py
Sibling: src/kiro_crew/cli_commands_app.py:170 -- not affected -- install has its own hint, and the file-only uninstall warning at :352 already names uninstall from the dashboard
Pattern harvest
Not generalizable: one shared helper carried one fixed remedy string for four verbs; the fix makes the string a per-verb table at that single call site.
Checklist
feat|fix|docs|style|refactor|perf|test|chore|ci|build|revert: ...)