RepoJournal
Vercel @vercel
125 wires and counting
$ status: archived — this briefing has stopped; the archive stays readable. browse live briefings →

$ status

wire 2026-05-13
stories 76
GitHub RSS Archive Reviews

© 2026 RepoJournal Home Showcase How it works Privacy

$ the-wire · showcase

VERCEL CLI SHIPS BREAKING CHANGES — LOGS NOW REQUIRE --FOLLOW FLAG

By RepoJournal · Filed · About Vercel · Composed from the cited sources · methodology

adriancooney tonypan2 bensabic lgrammel anaclumos
5 people shipped this
adriancooney @adriancooney 1 cited tonypan2 @tonypan2 1 cited bensabic @bensabic 1 cited lgrammel @lgrammel 1 cited anaclumos @anaclumos 1 cited

The Vercel CLI is tightening its logs behavior: `vercel logs` targeting a deployment now requires an explicit `--follow` flag to stream live output, treating all deployment arguments as historical filters by default.

This is a major breaking change [1] that forces developers to be intentional about live tailing versus historical inspection—a safety win for automation and scripts. The CLI also synchronized terminology across the stack [2], renaming "Connex client" to "Connex connector" to match the dashboard's updated feature naming. Over in the chat layer, adapters are getting more explicit—GitHub, Linear, and Slack SDKs are replacing the generic `.client` getter with SDK-specific names like `.octokit` and `.linearClient` [3], with deprecated aliases preserving backward compatibility. The AI SDK standardized on Start/End nomenclature [4], renaming `experimental_onEmbedFinish` to `experimental_onEmbedEnd`, while Google embedding models now support fileData for remote content [5]. On the security front, both vercel/workflow and vercel/ai removed the `pull_request_target` trigger from backport automation [6], eliminating a known privilege escalation vector in GitHub Actions. The SDK published v1.21.3 [7] with breaking response shape changes across projects, authentication, and deployments endpoints—review the Speakeasy changelog before consuming in downstream services.

Action items

References

  1. [1] cli/logs: Require explicit follow for deployments ↗ vercel/vercel
  2. [2] [cli] Rename "Connex client" to "Connex connector" in user-visible strings ↗ vercel/vercel
  3. [3] feat: rename adapter.client to adapter.octokit / adapter.linearClient / adapter.webClient ↗ vercel/chat
  4. [4] fix: rename onEmbedFinish to onEmbedEnd ↗ vercel/ai
  5. [5] feat(google): add fileData support to embedding model ↗ vercel/ai
  6. [6] Remove pull_request_target trigger from backport workflow (#1972) ↗ vercel/workflow
  7. [7] typescript - v1.21.3 - 2026-05-12 17:48:35 ↗ vercel/sdk

$ ls vercel/reviews/ # the step-back reads

the week in review Turborepo adds Go workspace graph, fixes hashing the month in review Workflow 4.8 overhauls delivery and sessions, Eve compacts traces, Turbopack cache defaults on
Full archive → Weekly reviews →

Quick answers

What shipped in Vercel on May 13, 2026?
The Vercel CLI is tightening its logs behavior: `vercel logs` targeting a deployment now requires an explicit `--follow` flag to stream live output, treating all deployment arguments as historical filters by default. In total, 30 commits, 35 pull requests, and 11 releases landed.
Who contributed to Vercel on May 13, 2026?
5 developers shipped this update, including adriancooney, tonypan2, bensabic, lgrammel, and anaclumos.
What were the notable Vercel updates?
cli/logs: Require explicit follow for deployments, [cli] Rename "Connex client" to "Connex connector" in user-visible strings, and feat: rename adapter.client to adapter.octokit / adapter.linearClient / adapter.webClient.
♥

$ log @vercel --recent

Turbopack GC gets a config key and the runtime sheds peer state 172 shipped xAI drops Chat Completions, eve breaks registry installs, Turbopack widens tree-shaking 155 shipped Next.js fixes cache warming env bug, eve gains audience hooks 167 shipped Turborepo patches Remix and Nuxt env hashing, shop ships Eve agent 10 shipped

$ log --all # elsewhere on the wire

Agent Security Quiet day in agent security as guard coverage reaches four clients Alignment & Post-training verl defaults Ascend RL mode, open-instruct unblocks SFT packing

$ ls showcases/

Safety Evals AI Red Teaming Agent Security Interpretability Alignment & Post-training

$ init your-desk

RepoJournal can write one of these for any repo you care about. Sign in with GitHub, pick your repos and sources, and your desk gets a wire like this one every morning.