Skip to content

Gallery — the artifacts you still need to review

You delegate work, and it comes back as visual artifacts — a design doc, a before/after mockup, a decision write-up, a report. Pasted into a chat, those get lost the moment the conversation scrolls. You end up hunting for the one you were supposed to look at.

dna sdlc gallery fixes that. It is the board-native index of every HtmlArtifact in a scope, grouped by the status of the work item that produced it. Because the index is generated from the board, it is always current — nothing to keep in sync by hand.

dna sdlc gallery                       # text panel — the review queue leads
dna sdlc gallery --html panel.html     # ONE self-contained page you can open
dna sdlc gallery --html panel.html --open   # …and open it in the browser
dna sdlc gallery --json                # the structured aggregate

The gallery is the sibling of dna sdlc digest. They answer two different "what happened while I was away" questions:

Shows Answers Reach for it when
dna sdlc digest events — timeline changes in a window "what happened?" you want the running story of the work
dna sdlc gallery artifacts — the HtmlArtifacts on the board "what visuals do I need to review?" you want to look at the deliverables

The digest tells you a Story reached review; the gallery hands you the mockup that Story produced, with a link, filed under Precisa de avaliação.

How an artifact lands in a bucket

The gallery walks every work item's outputs (produces[] ∪ the legacy back-refs — the same resolver the board uses) to find which work item produced each HtmlArtifact, then buckets the artifact by that work item's state:

  • 👀 Precisa de avaliação — produced by a Story in review, or by any work item with an open PR matched to it (from gh). This is your queue.
  • 🧭 Decisões — produced by an ADR. The visual that grounds a decision.
  • ✅ Shipado — produced by a work item in a terminal status (done, accepted, resolved, …). Done, kept for reference.
  • 📈 Em andamento — produced by a work item still being worked.
  • 📎 Sem work item — an orphan: an artifact nobody produced yet. Link it with dna sdlc produces add <WiKind>/<wi> HtmlArtifact/<name>.
🖼️  Gallery — dna-development  (6 HtmlArtifacts)
   artefatos visuais pra revisar, agrupados pelo status do work item

🧭 Decisões (1)
   • ha-dna-vs-agentframework  DNA vs Agent Framework — pivot portabilidade
       ← ADR/adr-dna-pivot-portability [accepted]
       🔗 https://claude.ai/code/artifact/34adbfd0-…

✅ Shipado (2)
   • ha-dna-dx-antes-depois  DNA DX — Agora → Depois
       ← Epic/e-dna-dx [done]
       🔗 https://claude.ai/code/artifact/8f9006ec-…

Getting artifacts onto the board

An HtmlArtifact stores an HTML page byte-faithfully as a first-class, linkable work-item output. Register one, stamp its hosted URL, and link it:

dna sdlc artifact create ha-my-mockup \
    --from ./mockup.html \
    --title "Checkout redesign" \
    --description "Before/after of the checkout flow" \
    --published-url "https://claude.ai/code/artifact/…"

dna sdlc produces add Story/s-checkout HtmlArtifact/ha-my-mockup

The --published-url is the canonical hosted location (e.g. a claude.ai artifact link). The gallery renders it as the clickable Abrir artifact ↗ on each card, so the panel points straight at the live page — not just the stored bytes.

An ADR can produce its artifact too (its decision-visualization), which is what lands it in the Decisões bucket:

dna sdlc produces add ADR/adr-my-decision HtmlArtifact/ha-decision-viz

The --html panel

--html <out> writes one self-contained HTML file — no CDN, no external assets, theme-aware (light/dark). Cards per artifact, a status chip, the producing work item, the published link, and the open PRs. It is the page you hand to whoever delegated: a board-native review surface they can open and regenerate any time to reflect the current board.

--open opens the generated file in your browser (writing a temp file if you didn't pass --html).