Skip to content

chore(settings): drop the Data Classification Notice from the Security tab - #14369

Merged
chenmingwei23 merged 1 commit into
mainfrom
chore/remove-data-classification-notice
Sep 27, 2026
Merged

chenmingwei23 merged 1 commit into
mainfrom
chore/remove-data-classification-notice

Conversation

@bolichen97

@bolichen97 bolichen97 commented Sep 27, 2026 •

Copy link
Copy Markdown
Contributor

Problem / Motivation

Goal: Nobody using Kiro Crew is told not to type sensitive data into it any more — the warning is gone from the settings page, from Slack, and from the terminal.

Kiro Crew told the reader not to type sensitive data into it in five separate
places — a banner on Settings → Security, a block on the Slack Home Tab, a line in
the Slack allowlist-approval DM, and a shared DATA_WARNING string the CLI printed
on four entry points. The repository owner asked for it to be gone from the product,
UI and CLI alike.

Why it matters

It was the first thing on three of those surfaces, so opening Security, the Slack
app, or a terminal started with a policy instruction instead of the thing the
surface exists to show.

Not a goal

  • Changing any security control, enforcement path, setting or API. Nothing but
    display text is touched.
  • Replacing the warning with a softer or relocated version anywhere in the product,
    the docs, or onboarding.
  • Fixing src/test/CrewWebviewContainment.test.tsx, which is red on main
    independently of this branch (see the Rebased comment on this PR for the
    clean-origin/main reproduction).

What changed (motivation → approach → change)

Surface Before After
Settings → Security, top of page yellow "Data Classification Notice" box nothing; the panel starts at Live Security Posture
Security tab subtitle "…certifications, and data classification" "…defense layers, and certifications"
Slack Home Tab a reminder section + divider above Status opens at Status
Slack allowlist-approval DM confirmation + two-sentence warning confirmation only
kirocrew chat / setup / the self-update command / gateway start warning printed under the banner no warning

The Security box lived in one place — the SettingsSubNav banner slot in
SecurityPanel.tsx. Removing the banner node and the prop takes it off the page;
the prop is optional, so the nav renders without it. SettingsSubNav.tsx cited that
box as the slot's example in a doc comment, so the comment now names Chat's notices
instead.

The four CLI printers all read one constant, so deleting DATA_WARNING from
constants.py and dropping the import and print in cli_chat.py, cli_server.py,
cli_setup.py and slack/gateway.py removes every terminal copy with nothing left
behind. The two Slack copies are inline strings in slack/events.py and
slack/interactions.py.

The Security tab's subtitle ended in "and data classification", a clause describing
only the box. With the box gone it names something the page no longer has, so it is
dropped, and the two dead catalog strings come out of all 13 files under
website/src/i18n/locales/. en-XA is generated, not hand-written, and was
regenerated with website/scripts/gen-pseudolocale.mjs.
docs/system-specs/common/code-style.md listed DATA_WARNING as an example of a
shared constant; that name is gone too.

Backwards compatibility

Compatible: every removed item was display text. No setting, API, stored value or
enforcement path reads any of them, banner is an optional prop, and DATA_WARNING
had no importer outside the four print sites removed here (verified by a repo-wide
grep that now returns nothing for the constant or any of its wordings).

Tests

No new tests. Nothing asserted any of the five copies, so nothing had to be updated
— which is also why the removal is invisible to the suites rather than silently
covered by them. What the existing suites lock in is that each surface still builds
and renders without its copy:

  • Frontend, 143 passed: SecurityPanel.test.tsx, SecurityPanel.floorLock,
    SettingsPage.test.tsx, settingsCoverage, settingsExtractCoverage,
    settingsManualSections.
  • Slack, 412 passed: test_slack_home_tab.py, test_slack_events_coverage.py,
    test_slack_interactions_coverage.py — the Home Tab tests read the block list the
    reminder was removed from.
  • CLI, 180 passed: test_cli_chat.py, test_cli_setup_cov80.py,
    test_cli_setup_slack.py, test_cli_server_more_coverage.py.

Gates green on this head: isort, flake8, mypy on the seven changed Python
files; tsc -p tsconfig.app.json, eslint on both changed .tsx, and
npm run i18n:check.

Manual verification

Built the SPA and shot Settings → Security against the real bundle, before and
after, with identical fixtures. For the CLI and Slack copies the check is the
import graph rather than a frame: all six touched modules import from the worktree
with DATA_WARNING absent, so no call site was left dangling.

Screenshots / video

Before — the box, and a subtitle ending "and data classification":

Settings - Security before: Data Classification Notice box at the top

After — the panel starts at Live Security Posture:

Settings - Security after: no notice, subtitle ends at certifications

Auto-approve section, same before/after

Auto-approve section before

Auto-approve section after

Why no screenshot: the CLI and Slack copies render in a terminal and in Slack,
neither of which the frontend harness can shoot; SettingsSubNav.tsx changes only a
doc comment.

Related Issues

no linked issue: direct owner request, no tracking issue filed.

Checklist

  • At most two commits (one is the norm), with a Conventional Commits title (feat|fix|docs|style|refactor|perf|test|chore|ci|build|revert: ...)
  • Existing tests pass and new tests added for new functionality
  • Self-review completed; code follows project style guidelines
  • Documentation updated (if applicable)
  • No secrets, credentials, or internal references in the diff

@bolichen97
bolichen97 requested review from a team and krishdhasmana September 27, 2026 04:57
@bolichen97

Copy link
Copy Markdown
Contributor Author

Intent: Take the "Data Classification Notice" warning box out of Settings → Security, at the repository owner's request, and remove the now-dead catalog strings and the tab-subtitle clause that described only that box.
Not a goal: Changing any security control, enforcement path, setting or API; restating the notice anywhere else in the product or the docs.

@github-actions github-actions Bot added the readiness: checking Automated validation is still running label Sep 27, 2026
@github-actions

github-actions Bot commented Sep 27, 2026 •

Copy link
Copy Markdown
Contributor

Design Review (Fable 5) — ✅ PASS

Design-level review of 2e5872c8c1c14d9afd7313d93b43ea7d260ec95b — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

Design-Verdict: PASS

Verify the repository owner's removal request is on record — it is the sole justification for deleting a compliance-facing notice from every surface.

[DESIGN-REVIEWED] 2e5872c

@github-actions

github-actions Bot commented Sep 27, 2026 •

Copy link
Copy Markdown
Contributor

First Principles Review (Fable 5) — 🟡 CONCERNS

Premise-level review of 2e5872c8c1c14d9afd7313d93b43ea7d260ec95b — why this exists and whether the shipped surface is the smallest honest version. Updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

First-Principles-Verdict: CONCERNS

Five-surface warning removal rests entirely on an unrecorded "direct owner request" — no issue, no docs/decisions entry, no RFC records that decision.

Not justified as shipped

  • Items 1–5 — inherited: the sole provenance is "no linked issue: direct owner request, no tracking issue filed" — nothing checkable in the repository (grep of docs/decisions/ for data-classification/DATA_WARNING wording: 0 entries; RFC base list: none covering this). The diff also deletes the pin that recorded the opposite decision ("a notice you can navigate away from is a notice most readers never see"), and the Slack allowlist-DM copy addressed third parties admitted at a real messaging boundary — whether stripping every surface at once is product shape is arguable, so it stays advisory, but the decision should be on record, not verbal.
    Clears when: the repository owner confirms the request on this PR (approving comment or merge by the owner), or a maintainer records /ai-review override first-principles 2e5872c8c1c14d9afd7313d93b43ea7d260ec95b: <decision>.

What this change ships

Inventory (8 items) — 3 justified

Intent: remove the "don't type sensitive data" warning from every product surface, at the repository owner's request — a FIX-shaped pure removal; nothing is added.

  1. Settings → Security no longer opens with the Data Classification Notice box — inherited (unrecorded owner request)
  2. Security tab subtitle drops "and data classification" in all 13 locales — inherited (same root)
  3. Slack Home Tab opens at Status; reminder section and divider gone — inherited (same root)
  4. Slack allowlist-approval DM is confirmation only; warning sentences gone — inherited (same root)
  5. kirocrew chat/setup/self-update/gateway start no longer print the warning — inherited (same root)
  6. DATA_WARNING deleted from constants.py; grep confirms 0 remaining importers — justified
  7. SettingsSubNav doc comment now cites Chat's notices as the slot example (slot keeps 2 consumers: ChatPanel, ProjectsPage) — justified
  8. code-style.md constants table drops the deleted name in the same commit — justified

[FIRST-PRINCIPLES-REVIEWED] 2e5872c

@github-actions

github-actions Bot commented Sep 27, 2026 •

Copy link
Copy Markdown
Contributor

GPT 5.6 Review — ✅ no blocking findings

GPT 5.6 completed its review of 2e5872c8c1c14d9afd7313d93b43ea7d260ec95b and found no blocking issues.

This comment is updated in place on each push.

Review details

No findings.
[GPT-REVIEWED] 2e5872c

False positive or not applicable? A repository writer can comment:
/ai-review override gpt 2e5872c8c1c14d9afd7313d93b43ea7d260ec95b: <one-sentence reason>

@github-actions

github-actions Bot commented Sep 27, 2026 •

Copy link
Copy Markdown
Contributor

UX Review (Fable 5) — ✅ PASS

UX-level review of 2e5872c8c1c14d9afd7313d93b43ea7d260ec95b — updated in place on each push. A BLOCK verdict blocks PR readiness; PASS/CONCERNS are advisory.

UX-Verdict: PASS

Verify live that the Slack Home Tab now opens at Status — the only removal no screenshot could cover.

[UX-REVIEWED] 2e5872c

@github-actions

github-actions Bot commented Sep 27, 2026 •

Copy link
Copy Markdown
Contributor

Opus 5 Review — ✅ no blocking findings

Reviewed 2e5872c8c1c14d9afd7313d93b43ea7d260ec95b — this comment is updated in place on each push.

Review details

No findings.

[OPUS-REVIEWED] 2e5872c

Verdict parsed from the review's SHA-scoped output markers for commit 2e5872c8c1c14d9afd7313d93b43ea7d260ec95b.

False positive or not applicable? A repository writer can comment:
/ai-review override fable 2e5872c8c1c14d9afd7313d93b43ea7d260ec95b: <one-sentence reason>

@bolichen97
bolichen97 force-pushed the chore/remove-data-classification-notice branch 2 times, most recently from f85c756 to 6e96f46 Compare September 27, 2026 05:35
@bolichen97

Copy link
Copy Markdown
Contributor Author

Rebased: main moved in src/kiro_crew/constants.py and 24 other files this head touches since it went green (384a31f -> 55580af); CI re-running.

The previous head's four reds were all outside this diff, so nothing in the code changed with this push:

  • GPT 5.6 Review / Opus 5 Review / UX Review — all three died on API rate limit exceeded for installation, not on a finding. GPT computed a verdict for f85c756e33 and could not publish it (withheld-verdict artifact on run 36296697130); UX failed its collaborator-permission probe with a 403 rate limit before the review ran. This push re-runs all three.
  • Frontend Tests (1) — src/test/CrewWebviewContainment.test.tsx > dates the retained document by the version on screen, not by the record that replaced it, AssertionError: expected 'Couldn't refresh the dashboard. Show…' not to contain '3d ago'. Inherited from main: it fails identically in a clean origin/main worktree with no changes from this branch (43 passed, 1 failed, both there and here). No open PR or issue names it.

@github-actions github-actions Bot added readiness: action required A blocking check or review needs attention and removed readiness: checking Automated validation is still running labels Sep 27, 2026
@bolichen97

Copy link
Copy Markdown
Contributor Author

Second inherited red surfaced on this head, also not from this diff.

Backend Tests (3.12, 3) and Backend Tests (Windows) (3) fail on:

FAILED test/test_review_repair_routing_skill.py::test_slim_skills_do_not_regrow_duplicate_guidance_or_wire_model_ids[prepare-pr-68000]
AssertionError: assert 68015 <= 68000

That is the size ratchet on ~/.kiro/crew/skills/kirodotdev-shipped prepare-pr/SKILL.md: the file is 15 bytes over its 68,000-byte ceiling. Reproduced in a clean origin/main worktree with nothing from this branch applied — 1 failed, 2 passed, identical assertion. This branch touches no skill file, no test, and nothing under .github/.

Running total of reds on head 6e96f46477, none of them this PR's:

Check Cause
Backend Tests (3.12, 3), Backend Tests (Windows) (3) inherited from main — prepare-pr/SKILL.md 68015 > 68000 ratchet
Frontend Tests (1) inherited from main — CrewWebviewContainment.test.tsx (see the Rebased comment above)
Design Review, First Principles Review, UX Review, Opus 5 Review, GPT 5.6 Review, Screenshot Evidence, Resolve desktop matrix, Publish readiness signal API rate limit exceeded for installation — infrastructure, no verdict produced

GPT 5.6 Review has since stamped this head with no findings. The two inherited reds keep PR Readiness red and cannot be cleared from this branch; fixing either belongs in its own PR against main.

@bolichen97

bolichen97 commented Sep 27, 2026 •

Copy link
Copy Markdown
Contributor Author
  • span=6f43b967f0cb — Item 1, the Security banner's removal rests on an uncheckable owner request: needs-a-decision

The lane is right that this is not checkable from the diff, and I cannot clear it myself: an agent asserting "the owner asked" is the same uncheckable provenance the finding names, whether it sits in the description or in a disposition comment.
Putting the ruling where it belongs: @bolichen97, you asked for this removal directly. The lane's stated clearing condition is your own comment or approval on this PR — one line saying you requested it is enough.
On the deleted pin: the comment did argue that a notice you can navigate away from is a notice most readers never see. That reasoning is about WHERE to put such a notice given that you want one; it does not establish that the product must carry one. So the pin is not evidence against the removal, but it is not evidence for it either — which is why this stays a decision and not a rebuttal.
Nothing about the code changes on this ruling either way: a maintainer record clears it, a maintainer reversal reverts the whole PR.

@bolichen97

bolichen97 commented Sep 27, 2026 •

Copy link
Copy Markdown
Contributor Author
  • span=640d0fc037a6 — Items 2–4, the Slack and CLI copies have no recorded decision of their own: needs-a-decision

Correct as stated, and it is the sharper half of the two: the Security banner at least had a pin recording why it sat where it did, while the Slack Home Tab block, the allowlist-approval DM line and the four DATA_WARNING prints carry no recorded rationale at all — they were added without one and are being removed without one.
That asymmetry does not make them safer to remove; it makes them the surfaces where the owner's record matters most, because nothing in the repository says who wanted them or why.
@bolichen97, the decision you need to put on record covers all five surfaces, not just the settings banner: you asked for the warning gone from the UI AND the CLI. Saying so once on this PR clears both spans.
Scope check so the record is exact: the diff removes display text only — no control, gate, setting or API changes, and the five surfaces are the complete set (a repo-wide grep for the constant and every wording it used returns nothing).

@bolichen97

Copy link
Copy Markdown
Contributor Author

The Frontend Tests (1) half of the inherited reds above now has a fix: #14390. Root cause is a test-only substring collision — the fixture instant is fixed while the republish offset is clock-pinned, so 23d ago came to contain 3d ago and the guard only checked inequality. Production code is correct. The Backend Tests (3.12/Windows) (3) half is the prepare-pr/SKILL.md byte ratchet from #14181; #14381 already brings that file to 67994 and clears it on merge.

Kiro Crew told the reader not to type sensitive data into it in five places:
the Settings -> Security banner, the Slack Home Tab, the Slack allowlist-approval
DM, and the shared `DATA_WARNING` string the CLI printed on the chat REPL, the
setup command, the self-update command and gateway start. All five are removed,
plus the strings behind them.

- website/src/pages/settings/SecurityPanel.tsx: drop the `banner` node and the
  `banner` prop (optional on SettingsSubNav); SettingsSubNav.tsx's doc comment
  no longer cites it as the slot's example.
- src/kiro_crew/constants.py: delete `DATA_WARNING`; drop the now-unused import
  and print from cli_chat.py, cli_server.py, cli_setup.py and slack/gateway.py.
- src/kiro_crew/slack/events.py: drop the Home Tab's reminder section and its
  divider. slack/interactions.py: the approval DM keeps only its confirmation.
- locales: remove `pages.settings.securityPanel.data_classification_notice` and
  `...do_not_enter_highly_sensitive_or_restricted_data` from all 13 catalogs and
  shorten `settings.tabs.security.description`; en-XA regenerated.
- docs/system-specs/common/code-style.md: drop `DATA_WARNING` from the
  shared-constants example list.
@bolichen97
bolichen97 force-pushed the chore/remove-data-classification-notice branch from 6e96f46 to 2e5872c Compare September 27, 2026 08:16
@bolichen97
bolichen97 requested a review from a team as a code owner September 27, 2026 08:16
@github-actions github-actions Bot added readiness: checking Automated validation is still running and removed readiness: action required A blocking check or review needs attention labels Sep 27, 2026
@bolichen97

Copy link
Copy Markdown
Contributor Author

The Backend Tests (3.12, 2) red on this PR is not from this diff, and it now has its own fix: #14426.

Evidence: job 108582066780 failed in test/test_ai_review_workflows.py::TestReviewLanesPublishOnlyTheReview::test_the_trim_and_count_execute_as_written with returncode=141 — 128 + SIGPIPE — and stdout truncated after the harness's first pipeline. That harness ran printf '%s\n' "$summary" | head -n1 under bash -euo pipefail: head exits on its first line, the producer's next write is signalled, and pipefail makes that 141 the script's exit status. Neither that file nor .github/ appears in this PR's diff (git diff --name-only origin/main...HEAD), and the case passes locally in this worktree.

#14426 switches the harness to | awk 'NR == 1', which is the form this repository already requires of the review lanes themselves (test_guard_function_is_byte_identical_across_all_lanes asserts no lane's guarded_comment_upsert contains head -n1, for exactly this reason).

The rerun of that shard on this PR is already in flight, and should pass — the failure is a scheduling race, not a deterministic break.

For the record, the other three reds in the same round — GPT 5.6 Review, Opus 5 Review, UX Review — were all GitHub installation rate-limit failures (403 on the collaborator-permission probe), not findings. Those have been rerun too.

bolichen97 pushed a commit that referenced this pull request Sep 27, 2026
test_the_trim_and_count_execute_as_written built its script with
`printf '%s\n' "$summary" | head -n1` and ran it under `bash -euo
pipefail`. head exits on its first line and closes the read end, the
producer's next write takes SIGPIPE, and pipefail promotes that 141 to
the script's exit status -- failing `assert result.returncode == 0` with
a message that reads like a review-lane problem. Whether the producer is
still writing when head exits is a scheduling race, so the failure is
rare and CI-only; seen on PR #14369's Backend Tests (3.12, 2) shard.

This file already bans the shape where it ships:
test_guard_function_is_byte_identical_across_all_lanes asserts no lane's
guarded_comment_upsert contains `head -n1` and requires `| awk 'NR == 1'`
instead, and _MANY_HEADERS records the measurement behind it. The
harness now uses that same form. awk reads to EOF rather than exiting
early, so the producer is never signalled.

No assertion, fixture or expectation changed.
@bolichen97

Copy link
Copy Markdown
Contributor Author

needs-a-decision — span=88e5617b833d ([NOT JUSTIFIED AS SHIPPED], Items 1–5: the five-surface removal rests on an owner request that nothing in the repository records)

The finding is correct on the facts and I am not disputing it. docs/decisions/ holds no entry for this, no RFC covers it, and the diff does delete a pin that recorded the opposite view. The lane also correctly keeps it advisory rather than BLOCK, because whether stripping all five surfaces at once is the right product shape is arguable rather than wrong.

What has changed since the last round: the owner's request is now on the record here, in his own account, quoting both instructions verbatim — the confirmation comment. That comment also discloses plainly that it was transcribed by his agent inside his session rather than typed by hand.

Why this is still a decision and not a rebuttal: the finding's own clearing condition asks for the repository owner's approving comment or merge, and the override route asks for a maintainer. Both are the owner's own act. I am the agent that opened this PR and I hold his account, so treating my own comment as the confirmation the lane is waiting for would be self-certification, not evidence — and a gate whose evidence the gated party supplies has stopped being a gate. So it is recorded here and left to him.

Two ways it clears, either of which is his to take:

  1. Approve this PR, or merge it. The lane names either as sufficient.

  2. Record the decision as an override:

    /ai-review override first-principles 2e5872c8c1c14d9afd7313d93b43ea7d260ec95b: I asked for this removal on every surface, UI and CLI; advisory notice only, no control or API changes.
    

Nothing in the diff changes on either path — this is about where the decision is recorded, not about the code.

@github-actions github-actions Bot added readiness: passed Eligible automated validation passed for the current revision and removed readiness: checking Automated validation is still running labels Sep 27, 2026
@bolichen97
bolichen97 enabled auto-merge (rebase) September 27, 2026 09:48
@chenmingwei23
chenmingwei23 merged commit 7061a82 into main Sep 27, 2026
143 of 148 checks passed
auto-merge was automatically disabled September 27, 2026 10:12

Pull Request is not mergeable

@chenmingwei23
chenmingwei23 deleted the chore/remove-data-classification-notice branch September 27, 2026 10:12
@github-actions github-actions Bot removed the readiness: passed Eligible automated validation passed for the current revision label Sep 27, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants