Background
Gap analysis vs the deployed ara-hub viewer (https://www.agenticresearch.sh/ara/AmberLJC/ara-paperbench/artifacts/speedrun/nanogpt-speedrun).
Problem
Every block in the detail pane (.block cards, crates/ara-viewer/src/detail.rs) is always fully expanded. Table-heavy artifacts repeat large exhibits on many nodes — the nanogpt-speedrun artifact repeats a 21-row + 6-row table on ~10 nodes — making the detail pane a wall of tables with the reasoning content buried between them.
The hub wraps each block in <details>/<summary> with a right-aligned count tag (e.g. "result · 2"), so readers can collapse heavy blocks.
Proposal
Convert .block sections (evidence, result, built on, sources, typed fields) into collapsible sections, default open, with an item-count tag in the summary row (claims count, exhibits count, etc.). Persist collapse state per node at most — no global persistence needed.
Acceptance
Background
Gap analysis vs the deployed ara-hub viewer (https://www.agenticresearch.sh/ara/AmberLJC/ara-paperbench/artifacts/speedrun/nanogpt-speedrun).
Problem
Every block in the detail pane (
.blockcards,crates/ara-viewer/src/detail.rs) is always fully expanded. Table-heavy artifacts repeat large exhibits on many nodes — the nanogpt-speedrun artifact repeats a 21-row + 6-row table on ~10 nodes — making the detail pane a wall of tables with the reasoning content buried between them.The hub wraps each block in
<details>/<summary>with a right-aligned count tag (e.g. "result · 2"), so readers can collapse heavy blocks.Proposal
Convert
.blocksections (evidence, result, built on, sources, typed fields) into collapsible sections, default open, with an item-count tag in the summary row (claims count, exhibits count, etc.). Persist collapse state per node at most — no global persistence needed.Acceptance
<details>or equivalent ARIA)