fix(site): complete localized legal page depth

[channprj]

Summary

• close the remaining independent-review M-001 rows for Korean, Japanese, and Chinese Terms pages
• apply the >=500 visible-character contract to all 4 locales × About/Contact/Privacy/Terms pages
• expand legal scope, license, unofficial status, user responsibility, warranty, and change guidance without changing layout

Verification

• Node 22.22.0 npm test: Vitest 53 passed, Node 18 passed, typecheck/build passed
• Terms visible text: ko 1,432; en 695; jp 1,296; cn 1,111
• git diff --check

Review context

The original trust-anchor requirement named About, Contact, and Privacy, which were already compliant. This follow-up also closes the first independent review’s broader legal-page interpretation so the M-001 table is fully remediated.

[Copilot]

🟢 Approval recommended

Changes are limited to static content expansion plus a straightforward test update, with verification indicating the build/test suite passes.

Pull request overview

This PR completes the “trust-anchor” / legal-page substantiiveness contract by extending coverage to the localized Terms pages and expanding Terms copy for Korean, Japanese, and Chinese to meet the >=500 visible-character requirement while keeping the existing layout/content structure.

Changes:

• Extend the “trust anchors” build test to include terms across all locale prefixes.
• Expand localized Terms content (ko/jp/cn) to cover license scope, unofficial status, user responsibility, warranty/limitations, and change guidance.

File summaries

File	Description
site/test/build.test.mjs	Adds terms to the list of trust-anchor pages validated for substantive visible text.
site/app/content/legacy-content.json	Expands localized Terms page content for ko/jp/cn to satisfy the substantiiveness/legal guidance requirements.

Review details

• Files reviewed: 2/2 changed files
• Comments generated: 0
• Review effort level: Lite

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[chatgpt-codex-connector]

💡 Codex Review

kmsg/Sources/kmsg/KakaoTalk/ChatWindowResolver.swift

Lines 135 to 137 in f9da988

	if let existingWindow = findMatchingChatWindow(in: kakao.windows, query: query) {
	standardizeReadableWindow(existingWindow, label: "existing chat window")
	return ChatWindowResolution(window: existingWindow, method: .existingWindow)

Disambiguate chat IDs before reusing existing windows

When multiple chats have the same display name, this name-only fast path can return any matching open window before openChatListRow gets a chance to match the requested synthetic ID. Consequently, send --chat-id can deliver a message to the wrong duplicate chat and read --chat-id can return the wrong transcript; only reuse an existing window when the registry proves the name is unique, otherwise resolve the exact chat-list row.

---

kmsg/Sources/kmsg/Commands/WatchCommand.swift

Lines 338 to 341 in f9da988

	guard let logicalTimestamp = message.logicalTimestamp else {
	return false
	}
	return logicalTimestamp > watchStartedAt

Preserve messages sent during the watch start minute

KakaoTalkTranscriptReader.logicalTimestamp constructs timestamps at minute precision, while watchStartedAt includes seconds. A message arriving after watch starts but within that same minute therefore has a timestamp earlier than watchStartedAt and is permanently filtered out, making a new watch silently miss messages until the next minute; startup history should be separated using the stabilized baseline rather than a second-granularity comparison against minute-only data.

---

kmsg/Sources/kmsg/Commands/SendImageCommand.swift

Lines 109 to 112 in f9da988

	} else {
	runner.log("send-image: confirmation sheet not observed; allowing direct-send path")
	Thread.sleep(forTimeInterval: 0.7)
	}

Verify direct image sends before reporting success

If the target window fails to focus, Cmd-V is not accepted, or the confirmation sheet takes longer than 1.5 seconds to appear, this branch merely sleeps and then prints the success message and may close the window. Those cases are indistinguishable here from a genuine sheetless direct send, so automation receives a false successful-delivery result; verify a transcript/UI state change before taking this path as success.

---

kmsg/Sources/kmsg/Auth/KakaoTalkAuthenticator.swift

Line 136 in f9da988

print("Attempting KakaoTalk login...")

Keep authentication progress out of structured stdout

When a logged-out session is recovered while running read --json or chats --json, this line prefixes the eventual JSON document with plain text; the credential prompts and keyboard fallback have the same output-channel behavior. JSON consumers can no longer parse the successful result, and the native MCP server reports INVALID_JSON_OUTPUT, so authentication progress and prompts need to use stderr or be suppressed for structured invocations.

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".