Who contributed to Vercel on July 31, 2026?

13 developers shipped this update, including anthonyshew, sampoder, acdlite, Zack Tanner, Marcos Hernanz, lubieowoce, Shalabh Chaturvedi, and TooTallNate, and 5 more.

What were the notable Vercel updates?

fix: Fall back to polling on macOS, refactor: Route MFE eligibility by manifest, and refactor: Route package consumers by manifest.

@vercel

Next.js + the Vercel platform - frontend infrastructure for the web

github ↗

Pick a date

Topics: JavaScript Full archive →

The Wire · Showcase

TURBOREPO RIPS OUT TOOLCHAIN DISPATCH, NEXT.JS STABILIZES RSC HYDRATION

By RepoJournal · Filed 06:56 UTC on July 31, 2026 · About Vercel

13 people shipped this

anthonyshew @anthonyshew 5 cited

acdlite @acdlite 1 cited

Zack Tanner @ztanner 1 cited

Marcos Hernanz @marcoshernanz 1 cited

Shalabh Chaturvedi @shalabhc 1 cited

alangenfeld @alangenfeld 1 cited

github-actions[bot] @github-actions[bot] 1 cited

Alex Grover @alex-grover 1 cited

CamdenA21 @CamdenA21 1 cited

sampoder @sampoder

lubieowoce @lubieowoce

TooTallNate @TooTallNate

github-actions @github-actions

Anthony Shew landed five major refactors removing ecosystem-provenance routing from Turborepo's task system, while Next.js shipped fixes for static fallback hydration and killed obsolete render options.

Turborepo closed a foundational architectural gap this cycle. Shew's five-PR stack removes `ToolchainId` runtime dispatch entirely, replacing it with explicit contract domains for task behavior, MFE routing, and package consumer selection [1] [2] [3] [4] [5]. The macOS file watcher now falls back to content-aware polling on startup failures instead of crashing outright [1], preserving native performance on Linux and Windows. Over in Next.js, Zack Tanner removed the now-unused `shouldWaitOnAllReady` render option [7] that became dead code after recent static-generation work. Andrew Clark fixed a hydration bug where static fallback shells were incorrectly flagging the head element as partial [6]. Marcos Hernanz optimized Turbo Task locking by skipping reader locks for immutable reads, reducing contention in the incremental system [8]. Workflow added compute instance tracking to the event log so observability can distinguish whether concurrent steps ran on the same warm instance [10], and the benchmarking work split STSO metrics to separate inline gaps from queue-hop overhead [9]. V0 SDK shipped v3.0.0 with a full migration to API v2, including sandbox-backed agents and multitenancy support [11] [12]. Chat added native `DateInput` and `NumberInput` modal children so bots can leverage platform primitives like Slack's datepicker instead of text validation [13].

Action items

→ Review Turborepo task contract changes before next monorepo deploy vercel/turborepo [plan]
→ Test macOS file watcher behavior in development after upgrading Turborepo vercel/turborepo [plan]
→ Migrate v0 SDK integrations to API v2 before v3 becomes stable vercel/v0-sdk [plan]

References

[1] fix: Fall back to polling on macOS ↗ vercel/turborepo
[2] refactor: Route MFE eligibility by manifest ↗ vercel/turborepo
[3] refactor: Route package consumers by manifest ↗ vercel/turborepo
[4] refactor: Route task behavior through contract domains ↗ vercel/turborepo
[5] refactor: Remove ToolchainId runtime dispatch ↗ vercel/turborepo
[6] Fix `isHeadPartial` when hydrating from a static fallback shell ↗ vercel/next.js
[7] Remove obsolete shouldWaitOnAllReady render option (#96401) vercel/next.js
[8] Skip reader task locking for immutable Turbo Tasks reads (#96179) vercel/next.js
[9] [benchmarks] Split STSO by inline vs queue-hop steps, add distribution diffs vs main (#3213) vercel/workflow
[10] feat(world): persist the compute instance that ran each step attempt ↗ vercel/workflow
[11] v3.0.0 ↗ vercel/v0-sdk
[12] Update SDK to API v2 (#152) vercel/v0-sdk
[13] feat(modals): add DateInput and NumberInput modal children ↗ vercel/chat

Quick answers

What shipped in Vercel on July 31, 2026?: Anthony Shew landed five major refactors removing ecosystem-provenance routing from Turborepo's task system, while Next.js shipped fixes for static fallback hydration and killed obsolete render options. In total, 57 commits, 58 pull requests, and 4 releases landed.
Who contributed to Vercel on July 31, 2026?: 13 developers shipped this update, including anthonyshew, sampoder, acdlite, Zack Tanner, Marcos Hernanz, lubieowoce, Shalabh Chaturvedi, and TooTallNate, and 5 more.
What were the notable Vercel updates?: fix: Fall back to polling on macOS, refactor: Route MFE eligibility by manifest, and refactor: Route package consumers by manifest.

NOTION COMMENTS LAND IN CHAT SDK ALONGSIDE MESSAGE LIFECYCLE OVERHAULS

Vercel's Chat adapter ecosystem just expanded to Notion, letting bots participate in comment threads with the same unified handler code that powers Slack, Linear, and GitHub.

+11

javascript 176 shipped

@vercel Aug 5

VERCEL-PY STANDARDIZES ON TYPESCRIPT SDK PAYLOADS, WORKFLOW ENGINE SHIPS QUICKJS VM

Python SDK workflows now speak the same wire format as TypeScript, eliminating serialization friction across the stack while the workflow engine ships inline step execution and WASM caching.

+10

javascript 116 shipped

@vercel Aug 4

WORKFLOW GETS QUICKJS ENGINE, EVE UNLOCKS DYNAMIC AGENTS, NEXT.JS 16.3.0 SHIPS

Vercel shipped three major feature releases across its core platforms overnight: Workflow gains an opt-in QuickJS WASM VM that cuts serialization overhead, Eve now routes to dynamic remote subagents, and Next.js 16.3.0 hardens turbopack's export analysis.

+14

javascript 147 shipped

@vercel Aug 3

EVE SHIPS SESSION COMPACTION. NEXT.JS TURNS ON TURBOPACK CACHE BY DEFAULT

Eve 0.29.5 lands manual session compaction and fixes TUI cancellation, while Next.js makes Turbopack's filesystem cache the new baseline for all builds.

javascript 17 shipped

Elsewhere on the wire

AI Agents Aug 6

LANGCHAIN HARDENS ANTHROPIC TOOL CALLING AND GIT CREDENTIALS AS OPENHANDS SHIPS 1.10.0

LangChain fixed a critical credential leak in evaluation runs while Anthropic tool schemas get proper fallback handling, as three major agent frameworks push releases across the ecosystem.

ai-agents 54 shipped

Local LLMs Aug 6

OLLAMA FIXES KV CACHE CORRUPTION, LLAMA.CPP HARDENS FILE SEARCH, VLLM KILLS MOE LEGACY CODE

Ollama patched a critical cache corruption bug that corrupts lazy snapshots during multi-request prefills, while llama.cpp hardened Windows junction traversal and vLLM removed 14 hardcoded MoE shards across quantization methods.