feat(playground): observable streaming chat + retrieval inspector

[hallelx2]

Rebuilds the Playground as a ChatGPT-style streaming experience over the engine's agentic tree-walk (/v1/answer/treewalk?stream=true), with full retrieval observability.

• Streaming: SSE proxy route pipes the tree-walk stream (demo corpus via internal key, or the user's own docs via session) to the browser; useTreewalkStream parses events into typed state.
• Split UI: chat thread (streaming answer + citation quotes) left; live Retrieval Inspector (structure-reading → per-hop reasoning → sections touched) + ScoreCard right.
• Scores: confidence, groundedness (verifies each citation quote occurs in the source), citations, pages read, hops, latency, cost, tokens.
• DocumentPicker: shared demo corpus vs. the user's documents.
• Removes the dead non-streaming keyword-matching action; keeps session persistence. Landing design tokens throughout.

Depends on engine PR (Flusher pass-through) for real SSE flushing.

[sourcery-ai]

Sorry @hallelx2, you have reached your weekly rate limit of 500000 diff characters.

Please try again later or upgrade to continue using Sourcery

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
vectorless-web	Ready	Preview, Comment	Jul 3, 2026 6:04pm

[coderabbitai]

Warning

Review limit reached

@hallelx2, you've reached your PR review limit, so we couldn't start this review.

Next review available in: 51 minutes

Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available.
You're only billed for reviews past your plan's rate limits ($0.25/file).

How can I continue?

After more reviews become available, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews.

How do review limits work?

CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability.

For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window.

Please refer docs for additional details.

Review details

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro

Run ID: 0819112c-5c7f-4730-b025-7340b98ad891

📥 Commits

Reviewing files that changed from the base of the PR and between 92f7c13 and c8c117c.

📒 Files selected for processing (11)

• apps/web/app/(dashboard)/dashboard/playground/page.tsx
• apps/web/app/api/dashboard/playground/documents/route.ts
• apps/web/app/api/dashboard/playground/grounding/route.ts
• apps/web/app/api/dashboard/playground/stream/route.ts
• apps/web/components/playground/ChatThread.tsx
• apps/web/components/playground/DocumentPicker.tsx
• apps/web/components/playground/RetrievalInspector.tsx
• apps/web/components/playground/ScoreCard.tsx
• apps/web/lib/actions/playground.ts
• apps/web/lib/cp-proxy.ts
• apps/web/lib/hooks/useTreewalkStream.ts

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch halleluyaholudele/observable-playground

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}