fix(web): address skill row review feedback

This commit is contained in:
Yichen Jiang
2026-08-06 17:21:13 +08:00
parent 690f7ae035
commit a667d2cd64
39 changed files with 326 additions and 142 deletions
@@ -14,16 +14,16 @@ The Web transcript renders `skill` calls through the generic fallback row, so a
The collapsed row uses a 16-pixel document-and-sparkle glyph and the Bash row's neutral hierarchy: tertiary glyph, secondary `Skill` title, caption separator, and tertiary skill name. Running, failed, and interrupted calls retain the transcript's shimmer, error dot and first-line summary, and warning dot semantics. A settled call expands through the whole summary row into a 260-pixel bounded `Instructions` card containing the exact durable result text; the existing trajectory `Inspect` handoff remains available below the card.
The row derives every visible value from the logged call/result slice. It reads the skill name from the recorded `name` argument and the instructions from durable result content, and never joins the current skill catalog for descriptions or provider metadata. The existing ACP `skill-load` recording is seeded through the real Web persistence and composition path for a keyless interaction and accessibility snapshot.
The row derives every visible value from the logged call/result slice. It reads the skill name from the recorded `name` argument and the instructions from durable result content, and never joins the current skill catalog for descriptions or provider metadata. Because a history page can contain a `tool/result` after its `tool/call` fell outside the window, the generic `HistoryEntry` envelope now carries the paired call's name, exact arguments JSON, and event time on result entries. The Host derives this transient annotation and the result render intent from the complete log; the runtime prefers an in-window call and otherwise materializes the same `ToolResultNode.call` and `callTime` from the annotation. An orphan result still has `call: null`, and a call-side render intent remains unavailable when its event is outside the page. The existing ACP `skill-load` recording is seeded through the real Web persistence and composition path for a keyless interaction and accessibility snapshot.
## Alternatives considered
- Keep the generic tool row and add only a `skill` color selector in `ui-conversation`. This leaves the redundant input envelope and generic expanded body in place, and makes the conversation package own a domain-specific visual rule.
- Add a new `skill` value to the host tool render-intent union. The keyed client slot already identifies this tool without changing the wire contract, so a new cross-boundary presentation value adds protocol and snapshot surface without enabling another consumer.
- Add a new `skill` value to the host tool render-intent union. The keyed client slot already identifies this tool; the cross-page fix belongs to the generic history pairing envelope used by every tool rather than a skill-specific presentation value.
- Export the conversation package's private `ToolRow` component for reuse. Client packages intentionally expose contracts rather than cross-package components; exporting it would couple independent feature packages to conversation implementation details.
## Consequences
`ui-skill` now depends on the public conversation toolview contract, locale and primitive packages, and React in addition to its reference-source dependencies. It owns a small copy of the disclosure-row chrome, so future global interaction changes must update this registrant alongside the Bash sample and conversation rows.
Cold replay stays deterministic when the installed skill catalog changes, and the transcript remains compact until instructions are explicitly expanded. The dedicated card intentionally shows the tool's complete framed output rather than extracting only `<skill_instructions>`, preserving exactly what reached the model and avoiding a second parser for the skill result format.
Cold replay stays deterministic across pagination and when the installed skill catalog changes, and the transcript remains compact until instructions are explicitly expanded. The generic pairing annotation also prevents other keyed tool rows and result presenters from changing identity at a page boundary without persisting duplicate data. The dedicated card intentionally shows the tool's complete framed output rather than extracting only `<skill_instructions>`, preserving exactly what reached the model and avoiding a second parser for the skill result format.