feat(chat): suppress streaming JSON for A2UI tool calls#245
Closed
blove wants to merge 4 commits into
Closed
Conversation
Card-shaped placeholder rendered in place of streaming tool-call JSON while an A2UI / json-render surface is being built. Three shimmer rows + 'Building UI…' status label. Themed via existing chat-tokens (separator color, surface-alt background) so it inherits A2UI theme overrides.
When the bound Message represents (or results from) a tool call to a known GenUI tool (default: generate_a2ui_schema / generate_json_render_spec), render <chat-genui-skeleton> in place of the streaming body. Skeleton persists after streaming since the actual rendered surface mounts via the existing agent.events\$ channel in a separate slot. The DEFAULT_GENUI_TOOL_NAMES set is overridable via the new genuiToolNames input, following the same convention as the LangGraph adapter's subagentToolNames.
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
…I sentinel
The original detection only fired post-streaming when tool_calls populates,
which is way too late: the user sees the entire stream of JSON args (or
the emit-phase a2ui_JSON payload) before the skeleton kicks in.
Three detection layers now cover the full streaming pipeline:
1a. Post-stream tool_calls[].name (unchanged).
1b. Live-stream OpenAI Responses-API content array — checks for any
`{ type: 'function_call', name: <genui tool> }` block. This shape
arrives on the FIRST stream chunk, before tool_calls populates.
1c. Emit-phase assistant message whose content starts with the A2UI
sentinel `---a2ui_JSON---` (or any prefix of it during the first
few tokens, so the skeleton fires from token 1).
2. Tool result message tagged with the GenUI tool name (unchanged).
Adds 4 new tests covering the function_call block, full sentinel,
partial sentinel, and a negative case that doesn't accidentally match
unrelated dash-prefixed content.
Verified live against the dev server with the "settings card" prompt:
skeleton now renders from the first token of both the tool-call and
emit phases.
blove
added a commit
that referenced
this pull request
May 11, 2026
Seven phases, single PR. Reuses chat-genui-skeleton from PR #245 (lifted into Task 1.1), classifier rename + patience fix (Phase 2), chat-tool-calls excludeToolNames filter (Phase 3), full chat-message revert (Phase 4), composition-level isGenuiTurn + skeleton + filter wiring (Phase 5), public-api export (Phase 6), controller-handled api-docs + PR open + #245 close (Phase 7). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
5 tasks
Contributor
Author
|
Closing in favor of #246, which moves the GenUI orchestration from chat-message (wrong layer — hid the surface mounting slot) to the chat composition (right layer — sibling of the surface branch). See spec at docs/superpowers/specs/2026-05-11-genui-streaming-architecture-design.md. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
When the LLM calls `generate_a2ui_schema` or `generate_json_render_spec`, the tool's args stream back as raw JSON in the assistant bubble — visible for ~3s before the rendered `` mounts and replaces it. This PR suppresses that JSON streaming and shows a `` instead.
Spec: `docs/superpowers/specs/2026-05-11-nav-v2-polish-palette-and-genui-suppression-design.md` (`GenUI stream suppression` section).
Test plan
🤖 Generated with Claude Code