feat(ai-elements): add Convex UI parts adapter

[HomenShum]

Summary

• add a typed Convex-to-AI-Elements adapter for persisted FastAgentPanel message parts
• expose one canonical ordered renderParts stream with exactly-one render ownership
• preserve text, reasoning, sources, files, normalized tools, domain passthrough, raw domain identity, and original ordering
• normalize malformed terminal tool states into AI SDK v5-compatible disjoint shapes
• keep aggregate projections convenience-only and keep persistent stream primitives out of this adapter

This is supporting infrastructure; runtime call-site adoption lands separately in UIMessageBubble.

Safety contract

• no primitive drives a persistent stream
• domain cards retain their raw payload identity
• unified and legacy tool call/result shapes collapse to one owner
• valid explicit terminal states retain precedence while malformed payloads are rebuilt
• null or blank error aliases do not manufacture a failure
• repeated unified snapshots cannot duplicate FIFO ownership
• tool output/error terminal states are disjoint
• exact ordering and exactly-once ownership are regression tested

Verification

• adapter: 33/33 passed
• full FastAgentPanel: 89 passed, 19 intentional skips
• root TypeScript: passed
• strict adapter TypeScript: passed
• design contract: 11/11 passed
• design lint: 0 high-severity violations
• production build/PWA: passed, 265 precache entries
• protected-preview authentication prerequisite: fix(ci): authenticate protected Tier B previews #522 merged
• independent post-fix release review: GO, no P0/P1 findings

No Convex production deploy and no admin merge bypass.

[vercel]

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

Project	Deployment	Actions	Updated (UTC)
nodebench-ai	Ready	Preview, Comment	Jul 14, 2026 11:24am

[github-actions]

PR size advisory

This PR adds 1108 lines of substantive change. CONTRIBUTING.md defines a soft limit of ~400 LOC.

If the PR is genuinely cohesive (e.g. an architecture map, a generated migration, a deletion of a dead module), no action is needed. Otherwise consider:

• splitting into 2-3 PRs along independent concerns
• pre-discussing the architecture change in a GitHub Discussion before merge

This is advisory — it does not block the merge.

[github-actions]

✅ Dogfood Visual QA Gate: PASSED

Check	Status
Screenshots	23 captured (pass)
Walkthrough	9 chapters (pass)
Key Frames	9 extracted (pass)
Scribe Steps	8 how-to steps (pass)
Build	success

Artifacts

Download the dogfood-evidence-f45fd40 artifact from the Actions tab for full screenshots, frames, and walkthrough video.

---

Generated by Dogfood QA Gate

[github-actions]

Tier B regression failed against preview

Preview URL: https://nodebench-jpck6ayip-hshum2018-gmailcoms-projects.vercel.app
Run: 29323883014

Likely a kit-parity DOM regression. Check the Playwright report artifact attached to the run for the failing selector + screenshot.

This check would have caught all 4 A9 sessionRows iterations as a single PR — that's why it exists.

[github-actions]

Demo: walkthrough of the surfaces this PR changed is available as a workflow artifact (pr-demo-521) at https://github.com/HomenShum/NodeBenchAI/actions/runs/29329076079