#Report Commands

#`report:translate`

Translate collected review JSON files to Korean.

npm run report:translate -- --my-app <owner> ...
node dist/translateReviews.js ...

#CLI Options

--my-app (required)
--registered-apps-path
--input-dir (default: data/{myAppId}/reviews)
--output-dir (default: data/{myAppId}/reviews-ko)
--target-lang (default: ko)
--provider (google-web | none, default google-web)
--overwrite (default false)
--concurrency (default 4)
--output text|json

#Example

npm run report:translate -- --my-app golden-horizon

#Output

data/{myAppId}/reviews-ko/{competitor}.json
data/{myAppId}/reviews-ko/.translation-cache.json

#`report:render-html`

Render actionable markdown report to a shared-viewer bundle JSON.

npm run report:render-html -- --my-app <owner> ...
npm run report:render-html -- --all
node dist/renderReportHtml.js ...
Default output is JSON bundle only. Legacy per-app HTML file is optional (--with-html).
HTML document title (<title> and H1) is tab-scoped: review tab uses {myAppId} 리뷰 ({myAppId} Review in English), and backlog tab uses {myAppId} 백로그 ({myAppId} Backlog in English).
Generated HTML top bar includes a Home button (/) at the top-left.
Filters are managed in a separate panel opened from the top bar Filter button.
On narrow screens, the filter panel opens as a bottom sheet.
On narrow screens, the notes panel also opens as a bottom sheet instead of a right drawer.
Search is shown as a 🔎 button by default and expands to the input field on tap/click.
The expanded search input fills remaining header space with flex layout to reduce awkward header wrapping.
Top controls are tab-scoped: Reviews and Backlog both show Filter and Notes.
In Backlog, Expand reviews is available inside the filter panel.
Top navigation state is synced to query parameters, so refresh keeps the same UI state:
- tab (review|backlog)
- q (search text)
- tags (multi-select tags, comma-separated)
- exclude (all|active|excluded)
- min100 (1 when 100+ chars is enabled)
- orig (1 when Original toggle is enabled)
- priority (all|must|should|could in Backlog)
- effort (all|high|medium|low in Backlog)
- page (Reviews pagination page number, 1-based; omitted on page 1)
The status row under the header is shown in both tabs:
- left: active filter chips (No filter when empty)
- right: selected/filtered count over total (Reviews x/y or Backlog x/y)
Backlog shows one unified backlog table across all apps (not per-app grouped sections).
In Backlog, identical backlog items are merged into a single row even when they come from different apps.
In Backlog, near-duplicate backlog rows (same normalized title/action) are also merged during backlog normalization.
In Backlog, default backlog sorting is: Priority (MUST > SHOULD > COULD) → Status (Not Started > In Progress > Done) → Effort (High > Medium > Low) → Review count (total count, high to low).
In Backlog, each row now includes Status (Not Started | In Progress | Done) and status can be changed inline.
Status uses color-coded states for quick scanning: Not Started (red), In Progress (yellow), Done (green).
Inline selectors in the backlog table are color-coded for both Priority (MUST/SHOULD/COULD) and Effort (High/Medium/Low).
The app list in each backlog row is rendered as a single-line text with ellipsis (...) when it overflows.
In Backlog, use the same filter panel UX as Reviews to filter rows by Priority / Effort.
In the Backlog table, there is no separate review-detail column; use the chevron button next to Review count to expand/collapse review rows.
Review count is displayed as active/total (for example, 2/8).
The total side of Review count is calculated as the number of unique reviews (reviewId-based dedupe), not raw quote line count.
In expanded review rows, the Korean sentence is shown by default (without KR: prefix), and See details reveals the review ID, metadata, and original text.
Expanded review rows render prioritized reviews only (top 8 max per backlog row).
Backlog themes are now derived dynamically from each app's review text (token-frequency heuristic), instead of using a fixed hardcoded theme list.
Synthetic backlog input now filters out low-signal reviews (e.g., short generic praise without request/issue) and keeps actionable reviews first.
Backlog action text is generated as a concrete checklist (up to 3 items) inferred from matched review patterns, not only a generic count sentence.
Backlog action text no longer appends count suffixes like (리뷰 N건) or (evidence N reviews); use the Review count column instead.
In Reviews, hashtag filter supports multi-select (#❤️, #Requests, #Satisfaction, #Dissatisfaction), and All tags clears tag filters.
In Reviews, app filter is available in the filter panel (All apps + per-app buttons).
In Reviews, state filter is tri-state: All / Active / Inactive (default: All).
In Reviews, app sections are ordered by latest review registration date first, then higher app rating.
In Reviews, app sections are always expanded; collapse/close interaction is not provided.
In Reviews, review cards are ordered by latest registration date first, with higher rating as the secondary sort.
In Reviews, cards are paginated (50 cards per page) with Prev/Next controls in the top status row; Prev scrolls to page bottom, and Next scrolls to the top of the review-card section.
In Reviews, bookmark controls are shown when no non-app filters are active (search/tag/state/100+ chars clear; app-only filter is allowed). Bookmark storage is scoped by app filter, so each app-filter view keeps its own bookmark. The bookmark update control is an icon button: it appears filled on the bookmarked page and outlined on other pages. The summary label shows progress (Bookmark {page}/{total}), and when bookmarked page reaches the last page it shows Bookmark {total}/{total} · Done. Go to bookmark jumps back to the saved bookmark page.
In Reviews, when bookmark is on the last page, active chips include Review status: Done (검토 상태: 완료).
In Reviews, you can toggle 100+ chars to focus on longer reviews.
Reset filters clears search/app/state/tag/length filters in one click.
In the filter panel, Reset all to inactive sets every review to Inactive in bulk and clears all hashtags.
The review filter sidebar/bottom sheet header shows filtered reviews / total reviews in real time.
Each app row in the Reviews tab also shows filtered reviews / total reviews on the right.
The context block below the title switches by tab, so each tab shows only relevant context.
In the Backlog tab context, a compact text summary is shown (backlog counts and hashtag-based priority rule) without stat/meta cards.
In the Reviews tab context block, a plain text summary is shown (app count, hashtag definition, active-state definition) without cards.
Review cards include #❤️ / #Requests / #Satisfaction / #Dissatisfaction hashtag toggles and Inactive/Active.
Curated quotes are shown in one Selected Reviews (Hashtag-based) section instead of separate category sections.
In review cards, Original text open/close uses a smooth expand/collapse animation.
Hashtags remain editable even when the card is Inactive (no click blocking by state).
Top-right controls include a Notes button; the right sidebar provides full note CRUD (create/select/edit/delete) with free-form titles and content.
The header action button ➕ 백로그 생성 is available in both Reviews and Backlog tabs.
In the notes panel, notes are managed from a dedicated list UI (not a backlog selector).
The notes editor has separate Title and Content fields per note.
Note Create persists immediately.
Note Save persists only the currently active note (Save button below the content field, or Ctrl/Cmd + S while the notes panel is open).
Note Delete removes the currently active note and persists immediately.
In Backlog, backlog rows are editable directly from the page:
- add/remove backlog items
- table columns are ordered as Priority → Backlog Item → Status → Effort → Review count (active/total) → Actions
- update each row's Priority / Status / Effort directly from inline selectors
- inline selectors use vivid red/yellow/green text colors with bold labels for clearer state separation (no state fill)
- Status labels use the same vivid red/yellow/green text-only style
- clicking a backlog table row opens the editor directly (except inline controls such as selectors/toggles/buttons)
- row action buttons are icon-only and pinned to the far right (Delete first, then Edit)
- deleting a backlog row uses a centered custom confirmation modal (not the browser confirm dialog)
- review selection is done in a centered modal with pagination and open/close animation (active reviews only)
- the review-picker modal keeps a fixed height regardless of result count, and only the list area scrolls
- editor summary shows both selected count and saved total count; if saved reviews are inactive, a guidance message is shown
- in the backlog editor Reviews list, each linked review can be toggled Active ↔ Inactive directly (persisted immediately to preview-state)
- setting a review to Inactive keeps the backlog linkage; use unlink (×) to remove it from evidence
- backlog editor body renders selected reviews as compact review text items (text only, no app name/ID chips)
- backlog editor header actions are Delete, Apply, and close (✕); Delete appears left of Apply
- in backlog editor, Apply saves immediately (persistent save), including Status
- report-table edits outside the backlog editor (row delete, inline priority/status/effort changes, quick-add from Reviews) are also auto-persisted
In Reviews, each review card has a single quick-add selector; selecting a backlog item toggles linkage immediately (attach when not linked, unlink when already linked).
In that quick-add selector, already linked backlog items are marked with ✓ (without priority labels), and the card shows a right-aligned linked-status text in the format (N linked: title, title).
The quick-add selector is rendered on a dedicated backlog row under the card action row (separate from tag/status controls).
When a review is added to backlog from Reviews, that review is automatically switched to Active.
The quick-add selector auto-refreshes backlog options without page reload: it revalidates on selector focus/open, on tab visibility restore, and periodically while the page is visible.
Reviews view is hydrated from full review datasets (data/{myAppId}/reviews-ko/*.json, fallback reviews/*.json) per app:
- preselected report quotes start as Active
- non-selected reviews are included as Inactive by default for manual curation
In preview mode, card states and notes are persisted to data/{myAppId}/reports/preview-state.json (both update immediately).
preview-state.json is treated as a full-state snapshot for review cards (not diff-only overrides).
preview-state.json now uses v4 schema (reviews.tags, reviews.excluded, notes.{id}.title, notes.{id}.content).
If data/{myAppId}/icon.png exists, HTML includes icon meta tags (icon, og:image, twitter:image) automatically.

#CLI Options

--my-app (required unless --all is set)
--all (default false): batch render all apps that have review JSON (data/{appId}/reviews-ko/*.json or fallback reviews/*.json)
--registered-apps-path
--input (optional): source file (.md or .json). If omitted, source apps are derived from raw review JSON.
--output (default: data/{myAppId}/reports/competitor-raw-actionable.ko.json)
--with-html (default false): also write legacy HTML file
--html-output (used with --with-html, default: data/{myAppId}/reports/competitor-raw-actionable.ko.html)
--all cannot be combined with --my-app, --input, --output, or --html-output

#Example

npm run report:render-html -- --my-app aurora-eos
npm run report:render-html -- --all

#Output

data/{myAppId}/reports/competitor-raw-actionable.ko.json (shared-viewer bundle payload)
data/{myAppId}/reports/backlog.ko.json (unified backlog items data including status; reviews are stored as scoped IDs in sourceToken::reviewId format)
Optional legacy output with --with-html: data/{myAppId}/reports/competitor-raw-actionable.ko.html

#`report:init-backlog`

Initialize backlog.ko.json per app (separate from preview-state initialization).

npm run report:init-backlog -- --my-app <owner> ...
npm run report:init-backlog -- --all
node dist/initReportBacklog.js ...
Keeps existing backlog by default and only initializes missing backlog files.
Uses report:render-html internally to create or normalize backlog data.

#CLI Options

--my-app (required unless --all is set)
--all (default false): initialize all renderable apps under data/
--registered-apps-path
--data-dir (default: data/)
--input (single-app only): optional report source (.md / .json) passed to report:render-html
--force (default false): regenerate backlog even if backlog.ko.json already exists
--dry-run (default false)
--all cannot be combined with --my-app or --input

#Example

npm run report:init-backlog -- --my-app aurora-eos
npm run report:init-backlog -- --my-app aurora-eos --force
npm run report:init-backlog -- --all

#Output

data/{myAppId}/reports/backlog.ko.json

#`report:init-state`

Initialize preview-state.json from report bundle defaults.

npm run report:init-state -- --my-app <owner> ...
npm run report:init-state -- --all
node dist/initReportState.js ...
Seeds all review states into data/{myAppId}/reports/preview-state.json using:
- reviewDefaults[reviewId].excluded (Active/Inactive default)
- reviewDefaults[reviewId].tags (default hashtags)
This command is intended for initial migration/reset. After initialization, manage card states in preview UI.
report:render-html writes report bundles (and optional legacy HTML) and does not reset preview-state.json.

#CLI Options

--my-app (required unless --all is set)
--all (default false): initialize all apps that have data/{appId}/reports/competitor-raw-actionable.ko.json
--registered-apps-path
--data-dir (default: data/)
--input (default: data/{myAppId}/reports/competitor-raw-actionable.ko.json)
--output (default: data/{myAppId}/reports/preview-state.json)
--keep-notes (default true): keep existing user notes while re-initializing review states
--all cannot be combined with --my-app, --input, or --output

#Example

npm run report:init-state -- --my-app aurora-eos
npm run report:init-state -- --all

#Output

data/{myAppId}/reports/preview-state.json

#`report:preview`

Run localhost preview server.

Dashboard mode: home lists app ids and generated report files (.md, .json, optional legacy .html)
Recommended flow: keep per-app data as JSON bundles and open /v/:appId (shared viewer).
Dashboard primary View Report link opens shared viewer route (/v/:appId)
/v/:appId reads competitor-raw-actionable.ko.json on the server and returns embedded html directly (no client-side bundle bootstrap)
Dashboard mode can display app icons when data/{appId}/icon.png exists
Report page header also uses data/{appId}/icon.png; when missing, UI falls back to appId text
Dashboard background fills the full viewport height (no abrupt cut when content is short).
Single-file mode: serve one legacy HTML file with --file
Serves preview state API for raw-review card management:
- GET /api/preview-state/:appId
- PUT /api/preview-state/:appId
- persistence file: data/{myAppId}/reports/preview-state.json
Serves backlog editing API:
- GET /api/backlog/:appId
- PUT /api/backlog/:appId
- request/response items[].evidenceReviewIds must use scoped IDs (sourceToken::reviewId)
- persistence file: data/{myAppId}/reports/backlog.ko.json
npm run report:preview -- [options]
node dist/previewReport.js ...

#CLI Options

--my-app (optional): dashboard filter
--registered-apps-path
--file: single-file mode path
--data-dir (default: data/)
--host (default: 127.0.0.1)
--port (default: 4173)

#Examples

npm run report:preview -- --port 4173
npm run report:preview -- --my-app aurora-eos --port 4173
npm run report:preview -- --file data/aurora-eos/reports/competitor-raw-actionable.ko.html --port 4173  # legacy mode

#Report Commands

#report:translate

#CLI Options

#Example

#Output

#report:render-html

#CLI Options

#Example

#Output

#report:init-backlog

#CLI Options

#Example

#Output

#report:init-state

#CLI Options

#Example

#Output

#report:preview

#CLI Options

#Examples

#`report:translate`

#`report:render-html`

#`report:init-backlog`

#`report:init-state`

#`report:preview`