mirror of
https://github.com/storytold/deckcraft.git
synced 2026-10-13 14:59:08 +00:00
Brandon ThomasandClaude Opus 5.5
a293b68258
lineInv, nonIsoscelesTrapezoid, pieWedge, the left-right / up-down / quad arrow callouts, circularArrow / leftCircularArrow / leftRightCircularArrow, the four curved arrows, swooshArrow, callout1-3, accentCallout1-3, borderCallout2-3, accentBorderCallout1-3, ellipseRibbon / ellipseRibbon2, leftRightRibbon, flowChartOfflineStorage, gear6 / gear9, funnel, the corner tabs (corner, square, plaque) and chartX / chartStar / chartPlus no longer fall back to plain rectangles. Each outline is our own construction in crates/geom/src/preset/extra.rs, fitted black-box against PowerPoint renders of synthetic probe decks at default, changed and extreme adjust values and several aspect ratios (outline within ~1pt on 300-480pt shapes). Tests cover catalog presence, hostile sizes/adjusts and measured outline points. Co-authored-by: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
25 KiB
25 KiB
DeckCraft vs Microsoft PowerPoint: parity assessment
Last reviewed: 2026-10-10 · Last updated: 2026-10-10 · Change: minor (hours to ~95% per audience added; ready recomputed from written weights, 56% → 55%; stage stays alpha; previous: major, full re-measure against PowerPoint for Mac 16.113.4; replaces the parity sections of ROADMAP.md) · Target: Microsoft PowerPoint for Mac 16.113 (Microsoft 365)
The authoritative answer to "how close is DeckCraft to PowerPoint, and how much work is left?". ROADMAP.md summarizes it; gaps.md itemizes every shortfall; parity-checklist.md is the row-by-row feature checklist behind the breadth number.
Headline
| Number | Value | Kind |
|---|---|---|
| Feature breadth (weighted checklist) | 79% over 191 features (P0 92%, P1 75%, P2 36%) | measured: cargo xtask parity formula over parity-checklist.md, rows re-scored 2026-10-10 |
| Menu-bar coverage | 59% (128 of 217 app-specific PowerPoint menu items have a DeckCraft command or UI label of the same name) | measured (heuristic lower bound): script matching the 2026-10-07 PowerPoint menu dump against the 226 engine command labels and the UI's strings; excludes the app, Window and Help menus and the subtitle language lists |
| Ready for real work | ~55% | estimated: weighted dimension table below (was ~56% until the features row was recomputed from written area weights) |
| Mainstream practitioner | ~53% | estimated: method below |
| Essentials user | ~62% | estimated: method below |
| Stage | alpha (beta is ~20 points and ~120–210 h away) | judgement against craftrules standards/progress-docs.md: ~55% is in the alpha band and all six core workflows pass the alpha gate |
| Remaining effort to beta | ~120–210 Opus 5.5 agent hours | estimated, itemized in roadmap.md |
| Remaining effort to full parity | ~365–650 Opus 5.5 agent hours | estimated, by dimension below |
The menu figure is lower than the checklist figure because PowerPoint's menu bar lists every chart
type, SmartArt category, action button and media source as its own item, and many DeckCraft
equivalents are one parameterized command (insert.chart {type}) that the label match cannot see.
Read it as a floor, not as a second breadth number.
Readiness by audience
| Audience | Ready | Opus 5.5 agent hours to ~95% | Work that dominates |
|---|---|---|---|
| Full target (ready for real work) | ~55% | 340–600 (~60% parallelizes: ~120–210 h wall clock with 3–4 agents) | Feature depth across all 20 areas, legacy .ppt/.odp and video export, localization in 11+ languages, GPU raster and hardware decode, AI features |
| Mainstream practitioner | ~53% | 150–250 (~50% parallelizes: ~70–120 h with 3 agents) | .pptx exchange (real-deck corpus, round-trip losses, EMF/SVG, embedded fonts) 45–75 h; depth in charts, animations, pictures, text, presenting, print/vector PDF 70–110 h; interaction (menu bar, Format Shape pane, font faces, dialogs) 25–40 h; stability 10–20 h |
| Essentials user | ~62% | 50–90 (~40% parallelizes: ~30–55 h with 2–3 agents) | Opening received decks (corpus subset, EMF, SVG icons) 25–45 h; Windows launch fallback and Close prompt 6–12 h; font faces, Tab, paste, print, show navigation 12–20 h; start screen, menu bar, palette 6–10 h |
Calibration is the same as for the full number (see Calibration): a fix is 0.5–1 h, a checklist row to PowerPoint depth 2–4 h, a subsystem 6–35 h. The sets are nested, so essentials ≤ mainstream ≤ full. The full number is an additive weighted sum over the dimension table; only mainstream and essentials use multiplicative discounts.
Target and how it was measured
- Target: Microsoft PowerPoint for Mac 16.113.4 (build 16.113.26100421), the Microsoft 365
build installed at
/Applications/Microsoft PowerPoint.app. PowerPoint for Windows has a few more features (Ink Replay, OLE editing, more export types, Hindi and Vietnamese UI); where it matters this document says so. - The installed app, black-box (AGENTS.md §2: names and listings only):
Info.plistCFBundleDocumentTypes: the formats PowerPoint opens and saves (13 roles; table in file-format-parity.md).Contents/Resources/*.lproj: 30 localizations plus Base (26 languages and 4 regional variants; table in localization-parity.md).- The menu-bar dump
plan/powerpoint/menu-tree.txt(491 lines, observed 2026-10-07, local and gitignored) and the ribbon/pane observations inplan/powerpoint/01-observed-ui.md.
- The code on
origin/mainatd556ff9: 226 engine commands (crates/engine/src/cmd/*), 48 transitions and 64 animation effects (crates/model/src/anim.rs), 504#[test]functions, the PPTX reader/writer (crates/pptx, 9.0k lines), 27 dialogs and 10 ribbon tabs (crates/ui-egui). - User reports: 24 GitHub issues opened 2026-10-08 to 10-10 (16 still open), the best evidence of depth we have.
- Not measured: rendering fidelity against PowerPoint (no reference renders or real-deck corpus exist yet), performance against PowerPoint, and any workflow end to end with a real user's deck.
By dimension
"Ready" is how close each dimension is to what a professional who presents with PowerPoint needs. The weights are what those users actually depend on, written down so the total can be recomputed.
| Dimension | Weight | Ready | Hours to full | Doc | Evidence |
|---|---|---|---|---|---|
| Features (breadth × depth) | 30% | ~62% | 140–240 | parity-checklist.md, animation-parity.md, slideshow-parity.md | Breadth 79% measured. Depth lags: 37 of 191 rows are partial, and users found depth bugs in rows scored done within two days (#53, #60, #69, #70, #74, #77, #78) |
| File formats | 20% | ~58% | 60–110 | file-format-parity.md | PPTX reads and writes every main part and keeps unknown objects; files open in PowerPoint without repair. But it has never met a corpus of real decks, EMF/WMF/SVG pictures don't decode, comments/media/rewind settings are lost on round trip, and there is no .ppt, .pptm, .odp, video or GIF export |
| UI/UX fidelity | 15% | ~60% | 30–50 | ui-parity.md | All 10 ribbon tabs and the contextual tabs exist; Format Shape pane, Animation Pane, dialogs and menus are thinner than PowerPoint's; no native macOS menu bar; user reports #17, #71, #73, #76 |
| Stability | 10% | ~60% | 10–20 | gaps.md | Never-crash rules, a panic guard round every command, PPTX fuzz/malformed tests. But startup crashes on some Windows GPUs (#40, #59) and Close discards edits without asking (#60) |
| Performance | 5% | ~55% | 10–20 | hardware-parity.md | Unmeasured against PowerPoint. Slides rasterize on the CPU (vello_cpu); Morph re-aligns outlines every frame (#67) |
| Hardware | 5% | ~40% | 25–45 | hardware-parity.md | GPU-composited window (wgpu), but CPU slide raster, software-only video decode, presenter view not placed on the second display, no pen pressure, no pinch zoom |
| Platforms | 5% | ~80% | 5–10 | ROADMAP.md | Ahead in reach: macOS, Windows x64/x86/arm64, Linux (5 formats), FreeBSD and the web, all from one release workflow. Held back by the Windows GPU crashes and Linux/Windows clipboard gaps |
| Localization | 5% | ~8% | 55–90 + native review | localization-parity.md | English only; no string catalog. PowerPoint for Mac ships 26 languages. Bidi and Arabic shaping in slide text exist |
| Ecosystem (templates, add-ins) | 3% | ~25% | 10–20 | gaps.md | 8 original themes vs PowerPoint's large template gallery; no add-ins or VBA (out of scope by policy) |
| AI features | 2% | ~10% | 20–40 (owner decision on models) | gaps.md | No Designer, Copilot, Speaker Coach, live subtitles or Translate (all Microsoft cloud services). Agents drive every command over MCP instead |
| Total | 100% | ~55% (Σ weight × ready = 55.3) | ~365–650 | ||
| Automation (✱, not weighted) | ahead | 0 | mcp.md, control-protocol.md | Every command is scriptable from the CLI, a JSON control channel and MCP, headless or against the running app; PowerPoint for Mac has AppleScript and VBA only |
About 60% of the hours parallelize across crates (formats, render, text, UI, media, localization
catalogs). Needs a human: native-speaker review of every translation, the owner's decision on AI
models and on legacy .ppt import, running PowerPoint to make reference renders and to confirm
round-tripped files open cleanly, and hardware we don't have (Windows GPUs from #40/#59, pens,
multi-display rigs).
Mainstream practitioner and essentials user
Two narrower readings of "ready", computed the same way in every Craft app (craftrules
standards/progress-docs.md). The stage still follows the full number above and the alpha gate.
Mainstream practitioner: ~53%
The typical professional who builds and presents decks every week: a consultant, teacher, sales or product person. Left out: add-ins, Copilot/Designer and other cloud AI, co-authoring and admin features, pens and recording hardware, and languages other than the user's own. Depth here is the depth of the parts of each area this user touches (common chart types, the effects people use, presenting rather than recording), so it's higher than the full-area numbers; exchange with PowerPoint is taken out of the areas and charged once, as a discount.
| Area (weekly use) | Weight | Depth | Evidence |
|---|---|---|---|
| Text: typing, fonts, paragraphs, bullets | 18 | 70 | Every Home-tab control; font weights (#69), Tab in text boxes (#78) |
| Slides, layouts, sections, views | 10 | 82 | Checklist 90% breadth in both areas |
| Masters, themes, backgrounds | 8 | 75 | Slide Master view, 8 themes; Notes/Handout masters rarely used |
| Shapes, format, arrange | 14 | 75 | Presets, connectors, smart guides; Format Shape pane thinner |
| Pictures: insert, crop, styles | 8 | 62 | Works from file; paste fails on Windows/Linux (#53); crop partial |
| Tables | 5 | 78 | Styles, merge/split, grow-to-fit |
| Charts: column, bar, line, pie with data | 6 | 58 | Chart Data dialog; elements partial |
| Transitions and common animations | 8 | 76 | 48/48 transitions; effects cover ~85% of use; pane has no timeline |
| Presenting: show, presenter view | 10 | 70 | Works; not auto-placed on the second display; no laser |
| Audio and video on slides | 3 | 62 | Plays; slow video in the show on Linux (#91) |
| Comments | 3 | 60 | Threads, resolve |
| Undo and clipboard | 4 | 75 | Unlimited undo; Paste Special partial |
| Open, save, PDF export | 3 | 65 | No print; raster PDF artwork |
| Weighted depth | 100 | 71.3 |
| Discount | Factor | Evidence |
|---|---|---|
| Interaction fidelity | ×0.92 | No native macOS menu bar, thinner Format Shape pane, Tab ends text entry (#78), font faces (#69), window chrome (#71) |
| Stability on real machines | ×0.92 | Startup crashes on some Windows GPUs (#40, #59); Close discards edits (#60, fix in PR #75); video stalls in the editor on Ubuntu (#91); arrows reported dead in the show on Linux Mint (#17) |
| File exchange with PowerPoint users | ×0.88 | Harsher than VectorCraft's ×0.92 because swapping .pptx is daily work for this user: never tested on real decks, EMF pictures blank, icons from PowerPoint don't come through (#56), comment threads, media options and rewind lost on save (#70, #74, #77) |
71.3 × 0.92 × 0.92 × 0.88 = 53.1 → ~53%. This is about the same as the full number, unlike
apps where mainstream comes out well above it. The dimensions mainstream leaves out
(localization, ecosystem, AI) weigh only 10% of DeckCraft's full number, and the biggest penalty,
.pptx exchange, hits every mainstream user.
Essentials user: ~62%
Someone who makes a few simple decks: open or start a deck, type, add pictures, present, export a PDF. Leaves out advanced options, pro workflows and exchange edge cases, plus everything the mainstream number leaves out.
| Core feature | Weight | Depth | Evidence |
|---|---|---|---|
| Start a deck from a theme (start screen) | 8 | 70 | Works; layout problems on the start screen (#17, #76) |
| Type titles and bullets, basic formatting | 20 | 78 | Works; #69, #78 |
| New slide, layout, reorder, duplicate | 12 | 88 | Works |
| Insert a picture | 10 | 65 | From file yes; paste fails on Windows/Linux (#53) |
| Basic shapes and text boxes | 8 | 85 | Works |
| Theme colours and background | 6 | 80 | Works |
| Transitions | 6 | 85 | Works |
| Simple animations | 5 | 80 | Works |
| Present full screen and move through slides | 10 | 85 | Works; #17 on Linux Mint |
| Undo / redo | 5 | 95 | Works |
| Save and reopen | 6 | 85 | Works; Close doesn't ask to save (#60) |
| Export PDF / print | 4 | 70 | PDF yes, print no |
| Weighted depth | 100 | 80.1 |
| Discount | Factor | Evidence |
|---|---|---|
| Launch and stability | ×0.90 | Some Windows PCs can't start the app (#40, #59), and the workaround is an environment variable; Close loses work (#60) |
| Discoverability and UI clarity | ×0.95 | Familiar ribbon and tabs; no native menu bar, start-screen glitches, a command palette whose default entry fails (#79) |
| Opening files people send them | ×0.90 | .pptx opens, but real decks are untested and EMF pictures and PowerPoint icons can come in blank (#56) |
80.1 × 0.90 × 0.95 × 0.90 = 61.6 → ~62%.
User evidence (GitHub, 2026-10-10, maintainers excluded)
- 38 issues from 22 reporters and 48 PRs by people other than the owner (bidi/Arabic, Morph, tables, logging, packaging, a Gentoo overlay; open fix PRs for #53, #64, #70, #72, #74, #77). Praise threads or "switched from PowerPoint" reports: 0 so far; the only comment threads are "same here" confirmations (#17, #33).
- Open issues (27): 13 on the core path (#91 video playback, #60 Close, #59 and #40 startup
crashes, #53 image paste, #78 Tab in text, #69 font faces, #70/#74/#77
.pptxround trip, #56 icons from PowerPoint, #17 show navigation and file menu, #76 start screen), 14 polish or niche (#81 and #64 browser exports, #79 palette, #72 browser comment timestamps, #73 icon consistency, #71 and #26 window chrome, #41 system theme, #32 text transparency, #18 desktop shortcut, #15 and #30 languages, #58 README link, #50 DMG signature). - Reading: people try DeckCraft for real decks and hit the core path. Nobody has reported replacing PowerPoint yet, which fits ~53–55%.
By feature area
Breadth is measured from the checklist; "ready" adds depth and is estimated. Hours are to PowerPoint parity for that area. Weight is how much a PowerPoint user relies on the area (points out of 110); the total "ready" is the weighted mean of the rows (6,865 / 110 = 62.4%), and it is the Features row of the dimension table. Automation and Platforms are scored elsewhere.
| Area | Weight | Breadth (measured) | Ready (est.) | Hours | Main shortfalls | Doc |
|---|---|---|---|---|---|---|
| Application shell | 5 | 71% | ~55% | 8–14 | No native macOS menu bar, thin start screen (templates, recent), one window per presentation missing | ui-parity.md |
| Files (commands) | 4 | 69% | ~50% | see file formats | Print, reuse slides, passwords, legacy and macro formats | file-format-parity.md |
| Slides and sections | 6 | 90% | ~85% | 3–5 | Zoom (summary/section/slide), copying slides between decks keeps formatting only partly | parity-checklist.md |
| Views | 4 | 90% | ~80% | 3–6 | Snap to grid and grid options, New Window | ui-parity.md |
| Masters and themes | 7 | 85% | ~70% | 8–14 | Notes and Handout masters, theme variants, .thmx, effect schemes |
parity-checklist.md |
| Shapes | 7 | 86% | ~75% | 8–14 | Edit points, text warp, WordArt depth | ui-parity.md |
| Format (fill, line, effects) | 6 | 81% | ~65% | 8–14 | Format Shape pane depth, bevel and 3-D, picture/texture fill options | ui-parity.md |
| Arrange | 3 | 100% | ~90% | 1–2 | Align to slide vs selected as a menu choice, More Rotation Options | ui-parity.md |
| Text | 15 | 80% | ~62% | 18–30 | Font weights beyond Regular/Bold/Italic (#69), Tab in text boxes (#78), IME, true vertical CJK, dictionaries, AutoCorrect, thesaurus, equations (PR #55 open) | localization-parity.md |
| Tables | 5 | 89% | ~75% | 3–6 | Draw table, borders and cell effects depth | parity-checklist.md |
| Charts | 7 | 65% | ~45% | 12–20 | Data grid editing, chart elements, the chartex types (waterfall, funnel, treemap, sunburst, histogram, box) |
parity-checklist.md |
| SmartArt | 4 | 33% | ~25% | 10–16 | Text pane, layout switching, the diagram data model on save | parity-checklist.md |
| Pictures | 7 | 48% | ~45% | 12–20 | SVG/EMF/WMF, clipboard paste on Windows/Linux (#53), artistic effects, remove background, compress | file-format-parity.md |
| Media | 4 | 67% | ~60% | 8–14 | WMA/WMV decode, record audio/screen, icons and 3-D models, playback options lost on PPTX round trip (#74) | file-format-parity.md |
| Transitions | 4 | 92% | ~80% | 4–8 | Transition sounds, 3-D transitions approximated in 2-D | animation-parity.md |
| Animations | 6 | 79% | ~65% | 10–16 | 64 of PowerPoint's ~190 effects, Animation Pane timeline, preview, rewind lost on save (#77) | animation-parity.md |
| Slide show and presenter view | 8 | 66% | ~60% | 10–16 | Second-display placement, laser pointer, recording, subtitles, ink depth | slideshow-parity.md |
| Review | 3 | 60% | ~45% | 8–14 | Modern comment threads on round trip (#70), compare, version history, real spelling dictionaries | file-format-parity.md |
| Draw (ink) | 2 | 33% | ~30% | 5–9 | Pen pressure, ink to shape/text/math, lasso | hardware-parity.md |
| Undo and clipboard | 3 | 88% | ~75% | 3–5 | Paste Special, image paste (#53) | ui-parity.md |
| Automation ✱ | — | 100% | ahead | 0 | mcp.md | |
| Platforms ✱ | — | 67% | ~80% | see platforms | ROADMAP.md | |
| Total (feature areas) | 110 | 79% | ~62% | ~140–240 |
What works today (evidence)
Carried from the 2026-10-07 ROADMAP status and updated with what has landed since.
- Document model: presentations, slides, masters and 11 standard layouts with placeholder inheritance (slide → layout → master → theme), sections, notes, comments, custom shows, header/footer fields, embedded media, unlimited undo on shared slide snapshots.
- Themes: 8 original themes with colour and font schemes, custom colours and fonts, backgrounds (solid, gradient, picture, pattern), slide size presets.
- Shapes: ~180 preset geometries with adjustment handles, text boxes, pictures (crop, corrections, animated GIFs that play and pause), tables with styles, charts (column, bar, line, area, pie, doughnut, scatter and more), basic SmartArt, groups, connectors glued to sites, freeform/curve/scribble, merge shapes, ink, action buttons, WordArt; flipped shapes keep their text readable.
- Text: in-place editing with caret, selection, keyboard and mouse; every Home-tab font and
paragraph control; bullets and numbering; levels; autofit; columns; vertical text; date fields in
their own format; hyperlinks drawn in the theme's hyperlink colour and followed in the show.
UAX #9 bidi and Arabic shaping with HarfRust, script font slots with per-cluster fallback,
justLowkashida, bidi caret and selection. - Editing: select, marquee, move, resize, rotate, adjust; smart guides; nudge; duplicate; z-order; group/ungroup; align/distribute; Format Painter; Selection pane; clipboard.
- Transitions and animations: 48 transitions including Morph (objects, words, characters, geometry outlines), 64 animation effects (entrance, emphasis, exit, motion paths, media), triggers, by-paragraph builds, Animation Painter, on a shared timeline engine.
- Slide show: full screen, keyboard/mouse navigation, blank screens, go-to-slide, pen, presenter window, rehearse timings, custom shows, Set Up Show, Reading View; links and shape actions followed on click.
- Views: Normal, Outline, Slide Sorter, Notes Page, Reading View, Slide Master; zoom; grayscale.
- Review: comments with replies and resolve, accessibility checker, a basic spelling list.
- Media: MP3, AAC/M4A, ALAC, WAV, AIFF, CAF, FLAC, Ogg Vorbis and Opus audio; H.264, HEVC, VP9 and AV1 video in MP4/MOV/WebM/MKV with pure-Rust decoders; poster frames, trim, fades, volume, loop, play across slides.
- Files:
.deckcraft(zip + JSON), PPTX import/export (also opens.potx/.ppsx), PDF export (slides, notes, handouts 1–9 per page, text layer, links, bookmarks), PNG/JPEG, outline text, AutoRecover. - Automation: 226 commands, every one reachable from
deckcraft-cli, the JSON control channel and MCP. - Platforms: signed releases for macOS (universal), Windows x64/x86/arm64, Linux (AppImage/deb/rpm/tar.gz/Flatpak, x86_64 and aarch64), FreeBSD and the web (v0.4.0, 2026-10-09).
Calibration: how the hours were derived
From this repo's own history (git log, gh pr list, the 2026-10-07 ROADMAP):
- The bulk build: engine, model, renderer, text, UI, show engine, PPTX, PDF, media and the release pipeline (today 117k lines of Rust, 59k of them media codecs and containers ported from FilmCraft) took about 30 wall-clock hours with up to three agents, i.e. roughly 60–90 agent hours, landing publicly between 2026-10-05 22:13 and 10-06 02:28. That is ~100 agent hours for today's 79% breadth / ~55% ready, with community PRs.
- Catalogue rows (connectors with glue, freeform tools, merge shapes, AutoRecover, the gradient editor) took 1–2 agent hours each to reach "done" (commits 10-06 00:49–01:34, parallel agents).
- Recent PRs (2026-10-08 to 10-10): Animation Painter +189 lines, animated GIF playback +393, hyperlink fixes +474, Morph geometry +781, UAX #9 bidi and Arabic shaping +1,902. At the ~150–300 tested lines per agent hour these suggest, that is 0.5–1 h, 1–3 h and 6–12 h respectively. PR open-to-merge times (often 10–30 h) measure the review queue, not the work, and were not used.
Scale used: a fix 0.5–1 h; a checklist row to presence 1–2 h; a row to PowerPoint depth 2–4 h; a subsystem (localization infrastructure, a real-deck corpus with fixes, EMF rendering) 6–35 h. The tail costs more per point than the bulk did: the remaining work is depth, fidelity and verification, which the first 100 hours mostly skipped.
Why the full-parity estimate rose from ~190 h (2026-10-07) to ~365–650 h: the old figure
covered the checklist rows, ribbon depth and performance only (~135 h of it is comparable to this
document's 140–240 h for features plus 30–50 h for UI). New in this measure: localization (55–90 h;
there is none yet), hardware (25–45 h), legacy and missing formats including .ppt, .odp, EMF/WMF
and video export, a real-deck PPTX corpus, AI features, and the depth bugs users found in features
scored done (16 open issues).
Revision history
| Date | Change | Summary |
|---|---|---|
| 2026-10-10 | minor | Added the readiness-by-audience table: each number gets its own hours to ~95% (full 340–600, mainstream 150–250, essentials 50–90). The full number stays an additive weighted sum; no change to the percentages |
| 2026-10-10 | minor | Added the mainstream practitioner (~53%) and essentials user (~62%) numbers with written weights, discounts and user evidence. Features dimension recomputed from written area weights (66% judgement → 62.4% weighted mean), which moves ready for real work from ~56% to ~55%; stage unchanged |
| 2026-10-10 | minor | Stage checked against the core-workflow gate (all six pass); stays alpha |
| 2026-10-10 | major | Created from ROADMAP.md's "How far from full parity" and "Status" sections; full re-measure against PowerPoint for Mac 16.113.4 (Info.plist, lproj, menu dump), 226 commands, 191-row checklist (79%), weighted dimension table (~56% ready), hours recalibrated from git history |