Who contributed to Vercel on May 20, 2026?

5 developers shipped this update, including lukesandberg, anthonyshew, mischnic, aayush-kapoor, and felixarntz.

What were the notable Vercel updates?

fix(next): move workflow-socket.json out of .next/cache to fix ECONNREFUSED, fix: Remove Vercel API mock unwrap usage, and fix: Remove api client unwrap usage (#12881).

@vercel

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

github ↗

Pick a date

Topics: JavaScript Full archive →

The Wire · Showcase

WORKFLOW SOCKET FIX STOPS CI CRASHES, TURBOREPO HARDENS AGAINST PANICS

By RepoJournal · Filed 06:00 UTC on May 20, 2026 · About Vercel

5 people shipped this

aayush-kapoor @aayush-kapoor 3 cited

lukesandberg @lukesandberg 1 cited

anthonyshew @anthonyshew 1 cited

mischnic @mischnic 1 cited

felixarntz @felixarntz 1 cited

The workflow team fixed a critical CI failure where stale socket files caused ECONNREFUSED crashes on Vercel, while turborepo removed unwrap() calls across five crates to eliminate panic paths in production.

The workflow socket fix [1] resolves a nasty bug where cached `.next/cache/workflow-socket.json` files persisted across Vercel CI builds, causing the webpack loader to connect to dead processes and crash with ECONNREFUSED. Moving the socket file to `.next/workflow-socket.json` (outside the preserved cache directory) eliminates the inter-build contamination entirely. Meanwhile, turborepo is systematically hardening its Rust codebase by removing unwrap() and expect() calls across the API mock [2], API client [3], microfrontends proxy [4], run summary [5], and task executor [6] crates. These are boundary paths where panics have no business happening; failures now surface as typed errors instead. On the Next.js side, rootParams shipped enabled by default [7], removing the config flag and opening the door for broader adoption once route handlers and Server Actions support lands. The Next.js team also added an instrumentation client inject option [8] to let next.config.js drive bootstrap code without touching user source files. AI SDK dropped missing READMEs for the otel and mcp packages [9] [10], and fixed a subtle Google provider bug where tool call args wouldn't form because the code assumed `willContinue` was always present when it could be absent [11]. Google's new managed agents feature is now supported in the SDK [12].

Action items

→ Merge workflow socket fix before next CI run to stop ECONNREFUSED crashes vercel/workflow [immediate]
→ Exit prerelease mode on workflow and ai when ready for stable release vercel/workflow [plan]
→ Test rootParams in your Next.js app now that it's enabled by default vercel/next.js [monitor]
→ Update to latest AI SDK if using Google provider with tool calls vercel/ai [plan]

References

[1] fix(next): move workflow-socket.json out of .next/cache to fix ECONNREFUSED ↗ vercel/workflow
[2] fix: Remove Vercel API mock unwrap usage ↗ vercel/turborepo
[3] fix: Remove api client unwrap usage (#12881) vercel/turborepo
[4] fix: Remove microfrontends proxy unwrap usage (#12880) vercel/turborepo
[5] fix: Remove run summary expect usage (#12879) vercel/turborepo
[6] fix: Remove task executor unwrap usage (#12878) vercel/turborepo
[7] Enable rootParams by default (#93863) vercel/next.js
[8] Add nextConfig.instrumentationClientInject ↗ vercel/next.js
[9] chore: add readme for @ai-sdk/otel ↗ vercel/ai
[10] chore: add readme for @ai-sdk/mcp ↗ vercel/ai
[11] fix(google): fix streaming tool call args ↗ vercel/ai
[12] feat(provider/google): add support for managed agents in the Interactions API ↗ vercel/ai

Quick answers

What shipped in Vercel on May 20, 2026?: The workflow team fixed a critical CI failure where stale socket files caused ECONNREFUSED crashes on Vercel, while turborepo removed unwrap() calls across five crates to eliminate panic paths in production. In total, 36 commits, 40 pull requests, and 12 releases landed.
Who contributed to Vercel on May 20, 2026?: 5 developers shipped this update, including lukesandberg, anthonyshew, mischnic, aayush-kapoor, and felixarntz.
What were the notable Vercel updates?: fix(next): move workflow-socket.json out of .next/cache to fix ECONNREFUSED, fix: Remove Vercel API mock unwrap usage, and fix: Remove api client unwrap usage (#12881).

NEXT.JS OPTIMIZES APP ROUTING ENTRY POINTS WHILE WORKFLOW CUTS STREAM LATENCY 20 PERCENT

Next.js extracted invariant app-page handlers into shared runtime modules while Workflow's stream write batching moves into core, eliminating per-token round trips on idle producers.

+11

javascript 126 shipped 2-min read

@vercel 1 day ago

TWO CRITICAL SECURITY PATCHES LAND IN WORKFLOW AND EVE OVERNIGHT

Next 16.2.11 and PostCSS 8.5.12 patches ship across workflow to seal CVE vulnerabilities, while Eve locks down workflow failure visibility.

javascript 153 shipped 1-min read

@vercel 2 days ago

WORKFLOW 5.0 BETA ARRIVES WITH HOOK TOKEN RETENTION, AI SDK FIXES GEMINI AND BEDROCK HALLUCINATIONS

Workflow shipped its most ambitious beta yet with experimental token retention, while the AI SDK shipped five critical fixes across Gemini, Bedrock, and dynamic tool validation in a single push.

javascript 132 shipped 2-min read

@vercel 3 days ago

NEXT.JS PATCHES FIVE SECURITY HOLES OVERNIGHT WHILE VERCEL CLI GAINS AI GATEWAY BUDGET CONTROLS

Five security advisories dropped across Next.js in the last 24 hours, ranging from origin validation on internal redirects to fetch cache key collisions that could leak cached responses between requests.

+12

javascript 145 shipped 2-min read

Elsewhere on the wire

AI Agents about 9 hours ago

CLAUDE OPUS 5 LANDS ACROSS THE STACK

The newest Anthropic model is now live in langchain, Cline, and llama-index, with native support for extended reasoning and 1M context windows.

ai-agents 28 shipped 1-min read

Local LLMs about 9 hours ago

OLLAMA LANDS LAGUNA SUPPORT AND CRUSHES MEMORY LEAKS WHILE SGLANG HITS V0.5.16 WITH CONFIDENCE-DRIVEN SPECULATIVE DECODING

Ollama shipped three critical performance and reliability fixes for Metal residency and concurrent access patterns, while SGL-Lang released 0.5.16 with a new speculative algorithm hitting 383.7 tok/s on DeepSeek-V4.