feat(autofix): Stream agent output via Markdown streaming variant

[priscilawebdev]

Summary

• Switch the live agent content in artifactLoadingDetails.tsx to <Markdown variant="streaming" />. Both block.message.content and block.message.thinking_content are produced incrementally during an autofix run, so the streaming variant — which re-keys tokens per chunk and applies an animation tailored to incremental output — is the right fit here.
• Other v3 cards (root cause, solution, previews) continue to render finalized artifacts with the default static variant; they're intentionally not touched.

Stacked on

Builds on top of #115879 (the <Markdown /> primitive adoption). Base the merge on that PR landing first.

Test plan

• Trigger an autofix run on an issue and watch the loading details panel — confirm streaming content renders progressively and the new animation feels right against Seer's actual token cadence.
• Compare with the previous (static) behavior to verify no regressions in layout, scrolling, or auto-scroll behavior inside the 200px-max-height container.
• Sanity check thinking_content rendering (use a run that surfaces thinking blocks).

[github-actions]

📊 Type Coverage Diff

✅ No new type safety issues introduced. Coverage: 93.56%

[sentry]

Bug: The prevTextLensRef in the <Markdown> component is not reset when content changes, causing the streaming animation to be skipped for new text.
_{Severity: LOW}

Suggested Fix

Reset prevTextLensRef.current to 0 when the content prop of the <Markdown> component changes. This can be achieved by adding an effect that triggers on content change or by adjusting the existing useLayoutEffect's dependency array to correctly handle the content update and reset the stored length.

Prompt for AI Agent

Review the code at the location below. A potential bug has been identified by an AI
agent. Verify if this is a real issue. If it is, propose a fix; if not, explain why it's
not valid.

Location: static/app/components/events/autofix/v3/artifactLoadingDetails.tsx#L40-L58

Potential issue: When a content block transitions from displaying `thinking_content` to
the final `content`, the `prevTextLensRef` within the `<Markdown>` component retains the
length of the previous text. This stale length is then used in a `useLayoutEffect` to
set an incorrect `data-skip` attribute for the new content. As a result, the animation
logic skips rendering the new characters, causing the text to appear instantly instead
of streaming in as intended.

_{Did we get this right? 👍 / 👎 to inform future reviews.}

[cursor]

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit db548af. Configure here.}

[cursor]

Stale animation state when switching thinking to content

Low Severity

When a block transitions from rendering thinking_content to content (e.g., content goes from "Thinking..." to a real value), both branches use the same key={index}. React reuses the Markdown instance, so the internal prevTextLensRef retains stale token text lengths from thinking_content. The streaming variant's useLayoutEffect then writes incorrect data-skip values onto the new content's DOM children, causing the decode animation to wrongly skip characters at the start of the actual content. Using a key that distinguishes the two branches (e.g., incorporating "content" vs "thinking") would give each a fresh Markdown instance with clean animation state.

Additional Locations (1)

• static/app/components/events/autofix/v3/artifactLoadingDetails.tsx#L50-L51

 

^{Reviewed by Cursor Bugbot for commit db548af. Configure here.}