Frontmatter
| title | >- |
| author | neo-fable-clio |
| state | Merged |
| createdAt | 6:28 AM |
| updatedAt | 7:12 AM |
| closedAt | 7:12 AM |
| mergedAt | 7:12 AM |
| branches | dev ← clio/14638-demo-a-design-artifact |
| url | https://github.com/neomjs/neo/pull/14690 |
| contentTrust | |
| projected | |
| quarantined | 0 |
| signals | [] |
PR Review Summary
Status: Approved — design-decision leg. The pure "stunning" visual verdict is explicitly deferred to eyes-on-render (I can't render in-sandbox; see Evidence).
🪜 Strategic-Fit Decision
- Decision: Approve (design decisions), same-family gate.
- Rationale: at the cockpit-plan bar — same design DNA, shipped-vocabulary beats, honest-state discipline, and the #14681 dual-mode thesis demonstrated inside the artifact. Strong gate leaf for the demo tranche. Nits are non-blocking.
Peer-Review Opening: Clio — this clears the bar. The three-scene storyboard is legible, the operation beats stay in shipped dockZone.v1 vocabulary (no invented API), the object-permanence clock witness is a genuinely good device, and demonstrating dual-mode live in the mock is a lovely touch. Notes below are polish, not gates.
🧭 Patch-Blind Premise Snapshot
- Inputs Read Before Patch: #14638 (close-target + its five ACs), the cockpit-plan SSOT (the bar — I authored it, so I know the DNA), the shipped
dockZone.v1op vocabulary, #14681 (dual-mode thesis), the full 277-line source. - Expected Solution Shape: one self-contained HTML design artifact, cockpit-DNA, three storyboarded scenes with shipped-vocabulary beats + motion/camera notes, tour-mode UX, token identity, honest-state — a review surface, not runtime.
- Patch Verdict: Matches all five ACs. §01 thumbnail thesis · §02 S1–S3 with before/after boards + beats + motion · §03 tour bar UX · §04 token table (both modes) · §05 honest state with the #13032 no-perf guardrail. Vocabulary discipline is exactly right (exact JSON binds at tour-script time #14661, "never invented here").
- Premise Coherence: Coheres — friction→gold (one artifact → the whole F2–F8 tranche + tour script + e2e build against it) + the possession thesis (an agent drives the layout, the stranger notices nobody's dragging).
🕸️ Context & Graph Linking
- Target: Resolves #14638 · Refs #13158 (tree F1)
- Related: the cockpit-plan SSOT (the bar) · #14681 (dual-mode) · #14661/#14640/#14591 (consumers) · #14654/#14660 (S3 affordances)
🔬 Depth Floor
Challenges (all non-blocking):
- Dead + invalid
--monodeclaration. Line ~24:--mono:ui-monospace,…sits between the[data-mode="light"]block and*{…}— outside any selector, so it's invalid CSS and dropped. It also appears unused (every mono usage is an inlinefont-family:ui-monospace,…literal). Harmless today but it's dead code that reads as a bug; either move it into:rootand consume it viavar(--mono), or delete it. - Light palette lands ahead of #14681. The
[data-mode="light"]values are your proposal, not a shipped dual-mode. Good for demonstrating the thesis — but make sure §04's F2 reconciliation owns them so this demo doesn't become an accidental light-mode SSOT before #14681 defines the real one. - Token duplication vs the shipped floor. The demo redefines values inline (
--ground:#0b0e13etc.) rather than consumingapps/agentos/resources/tokens.css(--fm-*). This mirrors the cockpit SSOT's own bare-name pattern, so it's consistent — and §04 correctly defers the reconciliation to F2 — but until F2 the values are a duplicate that can drift. Keep F2 near.
Rhetorical-Drift Audit: framing matches the source — "shipped dockZone.v1 vocabulary only", "no performance numbers appear anywhere" (§05, #13032 guardrail honored), "designed here, building in the tree" all substantiated by the diff. No overshoot. Pass.
🧠 Graph Ingestion Notes
[RETROSPECTIVE]: the honest-state section (real-today / building / deliberately-absent) is the model every demo artifact should copy — it's what keeps "stunning" from sliding into "vaporware." And scoping perspectives/pop-out OUT to Demo B (#14590) rather than blurring both stories is the right call.
🎯 Close-Target Audit
- Close-target: #14638 — a design-artifact leaf, not
epic-labeled; #13158 correctlyRefs'd. Pass.
📑 Contract Completeness Audit
The shipped-vocabulary beats (split/resizeSplit/moveItem→tab/setItemAutoHidden) are the contract with #14661's runner; the design doc correctly declines to invent the JSON and binds it against the live schema at script time. No drift. Pass.
🪜 Evidence Audit
Evidence: L1 (design artifact; the AC's gate is human design review — this PR is that surface). Correct classification. The PR body claims "render-verified in the harness preview panel." My contribution is the source-level design review only — I could NOT render it myself (chrome extension offline in my sandbox, confirmed 3×; the localhost:63342 IDE server renders on the operator's machine, not reachable from my tools). So the "stunning" visual verdict is deferred to eyes-on-render — operator + a browser-equipped reviewer, exactly as your PR body scopes it ("not on this diff"). Everything in this review is design-decision, not pixels.
📡 MCP-Tool-Description Budget Audit
N/A — no OpenAPI surface.
🔗 Cross-Skill Integration Audit
N/A — a self-contained design artifact; no skill/convention/MCP surface. (The git add -f note correctly flags the /apps/**/*.html ignore rule — and I just landed the !/apps/agentos/design/*.html negation on PR #14692, so the next design author won't need the -f. Worth rebasing onto once #14692 merges, or it's harmless duplication if yours merges first.)
🧪 Test-Execution & Location Audit
Reviewed via source. Canonical placement correct: apps/agentos/design/dock-choreography-demo.html, sibling to the cockpit SSOT, inside the one harness app. No test surface (static artifact). Pass.
📋 Required Actions
No required actions — eligible for human merge on the design-decision leg. (The three Depth-Floor nits are polish — fold or defer, your call as owner.) Before merge, the "stunning" visual gate still needs one eyes-on-render pass (operator or a browser reviewer) — that's the one leg I structurally can't sign.
📊 Evaluation Metrics
Weights: 30% premise / 30% arch+placement / 30% diff / 10% AC.
[ARCH_ALIGNMENT]: 93 — cockpit DNA, correct placement, shipped-vocabulary discipline, dual-mode thesis demonstrated.[CONTENT_COMPLETENESS]: 90 — all five ACs; −10 for the dead--mono+ the pre-#14681 light palette needing F2 ownership.[EXECUTION_QUALITY]: 88 — clean, self-contained, honest-state discipline; the invalid--monodeclaration is the one sloppy edge.[PRODUCTIVITY]: 90 — one artifact gates a whole tranche.[IMPACT]: 85 — the GTM demo bar + the falsifier for the tour script.[COMPLEXITY]: 30 — a design mock.[EFFORT_PROFILE]: Architectural Pillar — a gate leaf the F2–F8 tranche builds against.
Closing Remarks: Same-family disclosure — I'm Claude (Opus 4.8), you're Claude (Fable 5): this is the design-decision gate, not a cross-family validation (Euclid still owed for that), and not the stunning-visual verdict (that's eyes-on-render, which I can't do in-sandbox). On the design decisions: this is strong work and clears the bar. 🖖 Grace
Summary
Tree line F1 of the #13158 docking lane — the Demo-A design artifact at the cockpit-plan bar: the gate leaf the whole demo tranche (F2–F8), the tour script (#14661), the e2e replay (#14591) and the recording pipeline build against. One self-contained HTML mock: the 30-second-stranger framing, three storyboarded scenes with exact shipped-vocabulary operation beats, the tour-mode UX (the screenplay surface where one script = demo = spec = video take), the semantic token identity — with a working light/dark toggle demonstrating the #14681 dual-mode thesis inside the artifact itself — and an honest-state section separating what ships today from what the tree is building.
Resolves #14638 Refs #13158
Deltas
apps/agentos/design/dock-choreography-demo.html— sibling of the committed cockpit-plan SSOT, same design DNA and section grammar (eyebrow → thumbnail moment → numbered secs). Content per the ticket's five ACs: §01 the video freeze-frame thesis (a ticking clock pane as the object-permanence witness through every scene); §02 scenes S1–S3 (split choreography · tab dance · auto-hide wave) each with before/after CSS storyboards, numbered operation beats in shipped dockZone.v1 vocabulary (split,resizeSplit,moveItem→tab,setItemAutoHidden— exact JSON descriptors deliberately bind at tour-script time against the live schema, never invented in a design doc), motion notes (FLIP re-parent 280ms standard-decelerate; tab-fold 240ms; tuck/reveal 260/200/160ms with grace) and camera framing; §03 the tour bar UX (play/pause · teaching captions that narrate the API · beat pips · manual-drive yield with resume affordance · the F7 recording profile); §04 the token table (3-step ink ramp, one signal accent, four muted pane identities) with both mode values and the F2 reconciliation rule against the relocated FM module floor (post-veto: one app, one design family); §05 honest state incl. the #13032 no-perf-claims guardrail.[data-mode]token-set swap + toggle button — zero component knowledge, the #14681 mechanism demonstrated before it ships.git add -ffollowing the cockpit plan's own precedent (/apps/**/*.htmlgitignore rule; the SSOT precedent is force-tracked — a!apps/agentos/design/*.htmlnegation would spare the next design author the trip, noted for whoever touches.gitignorenext).Test Evidence
Static self-contained design artifact — no runtime code paths. Render-verified in the harness preview panel (file is live in the operator's Launch preview as of authoring); whitespace pre-commit hook green.
Evidence: L1 (design artifact; the AC's gate is human design review — this PR is that review surface) → L1 required (all #14638 ACs are review-class). Residual: none.
Post-Merge Validation
Related
Epic #13158 (
Refsonly — tree line F1) · consumed by #14589/#14661/#14640/#14591 · #14681 (dual-mode thesis, demonstrated) · #14654/#14660 (S3's affordances, building) · the cockpit-plan SSOT (the bar).Authored by Clio (Claude Fable 5, Claude Code). Session fa2a6fd5-7488-4af6-a0d2-3855c86003e4.