- 28 Sep, 2026 29 commits
-
-
next-js-bot[bot] authored
-
Hendrik Liebau authored
The grid page title used a random value whose string length varied. The RSC shell byte offset includes the title bytes. The test removes the title value before comparing two responses, but the `b:` row retains the original offset, so the equality assertion fails when the title lengths differ. The fixture now formats the value to 16 decimal places to keep the offset stable. [Flakiness metrics](https://app.datadoghq.com/ci/test/runs?query=test_level%3Atest%20%40git.repository.id%3A%22github.com%2Fvercel%2Fnext.js%22%20%40test.name%3A%22middleware-static-rewrite%20should%20revalidate%20the%20overview%20page%20without%20replacing%20it%20with%20a%20404%22%20%40test.status%3A%28%22fail%22%20OR%20pass%29&agg_m=count&agg_m_source=base&agg_t=count&fromUser=false&index=citest&start=1790020370174&end=1790625170174&paused=false) <img width="416" height="175" alt="Screenshot 2026-09-28 at 21 53 35" src="https://github.com/user-attachments/assets/6ca08128-6bd3-4c2e-aa41-37d2897425e0" />
-
Jimmy Miller authored
Run the loader build-dependency update assertion only with Turbopack. Webpack records `addBuildDependency` for build-cache invalidation but does not watch it for dev recompilation, which caused this assertion to fail in webpack CI. Replace the neighboring dependency-file test’s fixed delay with polling so it waits for the actual update.
-
Jarren authored
### What? Removes the three `CLAUDE.md` symlinks that point at `AGENTS.md` (repo root, `.github/`, and `turbopack/`), drops the "CLAUDE.md is a symlink" note from `AGENTS.md`, and removes the now-unused `CLAUDE.md` entry from `.alexignore`. ### Why? Claude Code now reads `AGENTS.md` directly, so the symlinks are redundant. Keeping a single instruction file avoids the two drifting apart. ### Notes - The `CLAUDE.md` fixtures under `evals/next-upgrade/evals/*` are intentionally untouched. They simulate user projects for the upgrade evals rather than repo instructions. - No changes to `create-next-app` or `@next/codemod`, which still generate `CLAUDE.md` for user projects.
-
Luke Sandberg authored
### What? Align `ExecuteContext` task accessors around their behavior when a task is absent: keep the frequently used `task`, use `try_task` for optional access, and `task_or_create` where storage may need to be materialized. ### Why? The three operations had inconsistent names despite sharing the same task-access mechanism. In addition, the diagnostic `get_task_name` path could create storage merely to name a missing task. ### How? Keep the shared private task-opening logic and its existing access policies; rename the public context methods and their callers. `get_task_name` now takes the optional path and returns `"unknown"` if the task is absent, without creating storage. The separate `Backend::get_or_create_task` API is unchanged. ### Verification - `cargo fmt --all -- --check` - `cargo test -p turbo-tasks-backend --test basic --test errors --test invalidation` (19 passed) - `git diff --check` <!-- fleet 8ec6f3d1-8af4-46e7-9a88-fd25742f533d --> --------- Co-authored-by:
vercel-fleet-prod[bot] <318278635+vercel-fleet-prod[bot]@users.noreply.github.com> Co-authored-by:
Luke Sandberg <210140+lukesandberg@users.noreply.github.com>
-
Jimmy Miller authored
Check one-row and wrapped layouts using the rendered header widths so the test remains valid as the displayed version changes.
-
Jimmy Miller authored
A star re-export barrel with imports but no top-level side effects is classified ModuleEvaluationIsSideEffectFree. follow_reexports_with_side_effects treats that as effectful and builds a SideEffectsModule that emits TURBOPACK_IMPORT(<eval part id>), but compute_side_effect_free_module_info classifies the same module as side-effect-free, so binding-usage prunes the Evaluation edge and the eval part never gets a module id. The build then panicked with 'ModuleId not found for ident: .../barrel.ts <internal part 0>' during middleware chunking. SideEffectsModule is designed so graph tree-shaking may trim these imports, so the consumer must tolerate a trimmed side effect. Add ModuleIdStrategy::try_get_id_from_module (returns None instead of erroring when an explicit id map lacks the module) and have chunk_item_content skip side effects with no id -- a pruned side-effect-free module has nothing to evaluate. Adds an e2e regression test (treeshake-mw).
-
Jimmy Miller authored
binding_source now carries the import's `with` clause (Option<Box<ObjectLit>>) so re-emitting the import in the consuming fragment keeps the same resolution semantics. Dropping it made transition imports (e.g. turbopack-transition) resolve to undefined at evaluation under tree-shaking.
-
Jimmy Miller authored
Turbopack: fix split_module correctness for app-router - Don't split the generated __nextjs-internal-proxy. client-reference proxy module. Its exports are all used, so splitting yields no benefit and duplicates its <exports> part in the module graph, tripping the duplicate-ident check in ModuleGraph::from_graphs. - Pass the original comments to each fragment so magic-comment annotations (turbopackChunkingType, turbopackExports) are still parsed. Fragment ASTs keep their original spans, so comments resolve at the same byte positions. Without this, references fall back to defaults (ChunkingType::Shared -> Parallel), breaking app-router layout-segment chunk groups.
-
Jimmy Miller authored
split_module skipped chunks whose body ended up empty, but the group index is used directly as a part id by outputs (Key::Export/Key::ModuleEvaluation), part_deps, and the PartId::Internal asserts embedded in the emitted chunk ASTs. Skipping shifted every later module's position and desynced those indices, causing out-of-bounds panics in split_module and wrong-fragment selection in part_of_module. Always push a module for every group; use a well-formed side-effect-free `export {}` placeholder when the body is empty. Update tree-shaker analyzer snapshots to include the placeholder parts. -
Hendrik Liebau authored
Startup preloading can initialize the built-in handler map before a configured handler finishes loading. A request that reaches `'use cache'` during the import can use the built-in handler instead of the configured handler. Next.js now waits for registration to finish before rendering that request. The server, route module, and export helper share the registration promise. Regression tests verify that the first request uses the configured handler with startup preloading enabled. The affected fixtures no longer disable preloading. Preloading also logs failed handler imports without an unhandled rejection.
-
Hendrik Liebau authored
On adapter deployments, a Draft Mode request for an unpublished URL incorrectly returns 404 instead of rendering the preview. This affects App Router pages with `dynamicParams = false` and Pages Router pages with `fallback: false` when the build did not generate that URL. An ordinary request to the same URL should still return 404. The adapter previously required both `__prerender_bypass` and `__next_preview_data` to route the request. Pages Router `setDraftMode()` and App Router `draftMode().enable()` set only `__prerender_bypass`. The adapter now routes requests whose bypass cookie matches the build's preview ID. Pages Router still validates legacy preview data when present and rejects malformed payloads. The Pages regression tests cover document and data requests, malformed preview data, and public cache isolation. The App Router draft test now runs on adapter deployments. These cases also run in dev and `next start`. The legacy builder deploy jobs skip them.
-
Will Binns-Smith authored
## Summary Follow-up to #98539 addressing [Luke’s `toSorted()` feedback](https://github.com/vercel/next.js/pull/98539#discussion_r4098899189), [signed Δ ordering feedback](https://github.com/vercel/next.js/pull/98539#discussion_r4098962512), and [locale-aware name tie-breaker feedback](https://github.com/vercel/next.js/pull/99369#discussion_r4125394575). The source table sorts individual rows and package aggregates by signed Δ in either direction; identical rows and groups remain last for every column. The separate impact-ranked route picker is unchanged, and table sorting uses `toSorted()` to preserve input arrays. Numeric ties now use the same locale-aware name ordering as the Name column. In compare mode the table defaults to **Changes**, showing added, removed, and changed sources; **All** opts into unchanged sources. The footer still reports full build totals. Single-build mode remains unfiltered because its rows are all compared to themselves. ## Verification - Focused filter checks: 6/6 passed (baseline 2/6). - Focused sorting checks: 12/12 passed (two new comparator cases failed before the fix). - `pnpm exec prettier --check apps/bundle-analyzer/components/diff-table.tsx` - `pnpm exec eslint --config eslint.config.mjs apps/bundle-analyzer/components/diff-table.tsx` - `pnpm exec tsc -p apps/bundle-analyzer/tsconfig.json --noEmit` <!-- NEXT_JS_LLM --> <!-- fleet d2af60a7-2332-4bf1-a425-0329992b44dd --> --------- Co-authored-by:
vercel-fleet-prod[bot] <318278635+vercel-fleet-prod[bot]@users.noreply.github.com> Co-authored-by:
Will Binns-Smith <755844+wbinnssmith@users.noreply.github.com>
-
Janka Uryga authored
`ensureStatic = "navigation"` was merged without rebasing, so the snapshots had some references to `unstable_{prefetch,navigation}` even though the APIs already dropped the `unstable_` prefix on canary. -
Will Binns-Smith authored
Virtualize source rows with `react-virtuoso` to avoid mounting the full table at once. Preserve sorting, filtering, grouped package expansion, sticky headers, totals, selection, and semantic table markup. Render an additional viewport around the visible range to avoid blank content during fast scrolling. <!-- NEXT_JS_LLM --> Co-authored-by:
vercel-fleet-prod[bot] <318278635+vercel-fleet-prod[bot]@users.noreply.github.com> Co-authored-by:
Will Binns-Smith <755844+wbinnssmith@users.noreply.github.com>
-
Janka Uryga authored
Debugging leftover from #99171.
-
Joseph authored
Remove unstable_ prefixes from navigation and prefetch in the entire codebase. [NAR-932](https://linear.app/vercel/issue/NAR-932/remove-prefixes-from-unstable-navigation-unstable-prefetch)
-
Jimmy Miller authored
Enables lazy imports for ssr. https://github.com/vercel/next.js/pull/97305 This was the old Pr for this. Tried to incorporate the feedback there. The changes here were from testing on v0/chat
-
Aurora Scharff authored
## Summary Align the Security Insight with the existing DevTools and Insight-card typography. The upgrade handoff remains agent-first, while the prompt is easier to scan with: - the standard Insight heading style - normal body-copy spacing and typography - a compact Agent prompt box with the existing Insight-style Copy prompt action - the exact paste-ready Markdown shown in the prompt body Adds a Storybook example for visual review and keeps the clipboard behavior covered by the existing test. ## Before and after | Before | After | | --- | --- | | <img width="446" height="733" alt="Previous Vulnerability Insight" src="https://github.com/user-attachments/assets/ae0bab2a-4549-4441-b27d-327ad49b16c5" /> | <img width="446" height="733" alt="image" src="https://github.com/user-attachments/assets/f37a4520-3516-4c9c-b529-9eb784be5b27" /> | ## Verification - Viewed the component in local Storybook at desktop and narrow widths. - Focused formatting and ESLint checks passed. - Integration test updated but not run; this draft was verified in Storybook without a full framework build. <!-- NEXT_JS_LLM -->
-
Janka Uryga authored
Closes NAR-930 Initial implementation of `ensureStatic = "navigation"` ### Behavior This version of `ensureStatic = "navigation"` makes all prerenders blocking. All params must be specified via `generateStaticParams`, similar to `output: "export"`. This is meant to ensure that we don't serve a fallback + a resume for a param that wasn't prerendered, which would violate the expectation that `ensureStatic = "navigation"` means no per-request compute, which is what we're trying to provide. Note that Client-only IO (like `use(io())` or `useSearchParams()`) is allowed, because it doesn't require a resume render -- we serve HTML with fallbacks and render the missing content in the browser. (this is existing behavior and not changed here) Earlier versions of this PR attempted to allow fallback shells if they managed to produce a complete RSC result, e.g. on a route with params that are only used via `useParams()`, where could use the same browser bailout trick as other client IO. This proved complex to implement correctly, so we'll revisit that in the future ### Prerender output changes We skip producing all fallback shells except the most generic one in `buildAppStaticPaths`. Initially, i wanted to remove all of them, but it seems like the deploy output is incorrect if we don't have segment data from a fallback prerender. We also error for incomplete/missing `generateStaticParams`, which makes it impossible to create a purely-fallback param. ### Validation This PR extends the Static Shell Validation logic to be more restrictive when `ensureStatic = "navigation"`. We can't allow _any_ usages of dynamic data, even when wrapped with suspense. The new logic is expressed in `trackDynamicAccessInStaticRoute` + `getDisallowedReasonsInStaticRoute` which we use both in dev and prod. we check whether the server prerender was partial (which is already an error) and then run static shell validation so we can report where the hole was. This has some drawbacks: - if we have both server and client holes we might misreport the client ones as dynamic data. I have some ideas for how to address this, but for now this is the best we can do. (see `mixed server and client holes` in `ensure-static-navigation.test.ts` for examples) - if the client got a dynamic data promise but didn't `use()` it during the prerender (maybe because it was just seeding a cache for use in the browser), we won't have a dynamic hole to point to. in that case we'll error with a generic error message instead (see `errorPattern.body.withoutLocation` in `ensure-static-navigation.test.ts`). i've added new error messages to match the constraints of `ensureStatic = "navigation"`. we no longer recommend `instant = false` or `suspense` for dynamic data, because that's not a valid fix. there's a new suggestion: "make your \`params\` static via `generateStaticParams\`". (note that the messages don't have the "pretty" errors with suggestion cards yet, just the string message. i will update the dev error logic in a follow-up. docs are needed as well) i've also done a bit of cleanup: `trackDynamicHoleInStaticShell` and `trackDynamicHoleInRuntimeShell` are now one function that takes a `kind` parameter, because they only differed on the error message variant they reported (runtime vs dynamic).
-
Joseph authored
Links to ISR'd pages need to have `prefetch={true}` set, plain links only prefetch an App Shell -
Aurora Scharff authored
## Summary - include experimental agent feedback in the interactive recommended defaults - ask about agent feedback only when someone chooses a non-default interactive setup, with **Yes** selected by default - generate `experimental.agentFeedback: true` when enabled and support explicit `--agent-feedback` and `--no-agent-feedback` flags - keep CI, `--yes`, and other non-interactive flows opted out unless `--agent-feedback` is provided - document and test the recommended and customized setup flows ## CLI experience Choosing the recommended defaults enables agent feedback without another prompt: ```text ? Would you like to use the recommended Next.js defaults? ❯ Yes, use recommended defaults - TypeScript, ESLint, No React Compiler, Tailwind CSS, No src/ directory, App Router, No Cache Components, AGENTS.md, Agent feedback No, customize settings - Choose your own preferences ``` Choosing a non-default setup asks: ```text ? Would you like to help improve Next.js by letting agents prepare anonymized feedback for your review as you code? (Disable anytime with `experimental.agentFeedback: false`.) › No / Yes ``` **Yes** is selected by default. ## Verification - `pnpm --filter=create-next-app build` - `HEADLESS=true pnpm --config.minimum-release-age=0 test-start-turbo test/production/create-next-app/prompts.test.ts -t "should use recommended defaults|should show reuse previous settings|should not prompt user for choice and use defaults if --yes is defined"` - `HEADLESS=true pnpm --config.minimum-release-age=0 test-start-turbo test/production/create-next-app/index.test.ts -t "should list both agent feedback flags in help|should enable agent feedback with --agent-feedback|should not print assumed defaults when all flags are provided"` - Next.js watch build and type checking completed with no errors - Prettier and ESLint passed <!-- NEXT_JS_LLM -->
-
Aurora Scharff authored
## Summary - make `next-dev-loop` open `agent-browser` headlessly for routine verification - keep one worktree-scoped restored `agent-browser` session - open the current URL in the coding harness's inline browser when the user asks to see the UI, while making clear that it is a separate browser context - relaunch the same `agent-browser` session with `--headed` only when login or existing browser state must carry over - update Cache Components adoption guidance to use the same explicit browser handoff choices ## Why Runtime browser verification is still needed for rendered behavior, navigation, browser errors, and React state that compilation and type checking cannot prove. That verification does not need to open a visible browser for every task. The coding harness may not be able to embed an existing `agent-browser` session. The instructions now distinguish an inline presentation handoff, which has separate cookies and storage, from a headed `agent-browser` handoff that preserves the restored session's state. ## Verification - Prettier passed for `skills/next-dev-loop/SKILL.md` and `skills/next-cache-components-adoption/SKILL.md` - verified the headless `agent-browser` flow and a separate coding-harness inline-browser handoff against a running Next.js app - not run: test suite (skill instructions only) <!-- NEXT_JS_LLM -->
-
Sebastian "Sebbie" Silbermann authored
The other test already use 4minutes. [This on is even on Turbopack barely below 60s](https://app.datadoghq.com/ci/test/runs?query=test_level%3Atest%20%40git.repository.id%3A%22github.com%2Fvercel%2Fnext.js%22%20%40test.name%3A%22Request%20Promises%20On%20Prerender%20Completion%20should%20reject%20request%20APIs%20after%20the%20prerender%20is%20complete%20when%20it%20finishes%20naturally%22%20%40test_session.name%3A%2Adeploy%2A&agg_m=count&agg_m_source=base&agg_t=count¤tTab=overview&eventStack=&fromUser=false&index=citest&start=1789888928703&end=1790493728703&paused=false) so worth doubling for every bundler anyway
-
Joseph authored
Found by our internal tool
-
Jiwon Choi authored
Stacked on #99291. ### Why? Let users revisit upgrade choices before the new agent starts. ### How? Use the existing CLI selector with a dim hint above each menu. Esc goes back to the previous choice, or cancels from the first prompt; Ctrl+C cancels from any prompt. Remove the Cancel rows. <!-- NEXT_JS_LLM -->
-
Jiwon Choi authored
Stacked on #99235. ### Why? Let users choose whether a new upgrade agent uses Auto permissions. ### How? Ask Yes/No after model and effort. No sets explicit approvals; CLIs without Auto use a supported approval mode. Existing agents keep their permissions. <!-- NEXT_JS_LLM --> Closes NAR-947
-
Jiwon Choi authored
Stacked on #99232. ### Why? Let users choose the model and effort for a new upgrade agent. ### How? Ask for model, then effort, in separate selectors and pass both to the CLI. Existing agents keep their settings. <!-- NEXT_JS_LLM --> Closes NAR-947
-
Jiwon Choi authored
### Why? Let users choose whether an upgrade runs in a separate worktree. ### How? Ask in the existing agent's single upgrade question or before launching a new agent, then pass the choice into the upgrade instructions. <!-- NEXT_JS_LLM --> Closes NAR-947
-
- 27 Sep, 2026 3 commits
-
-
next-js-bot[bot] authored
-
Jiwon Choi authored
Stacked on #99222. ### Why? Developers running `next dev` with `agenticAutoUpgrade` enabled need a browser reminder when their installed Next.js version is affected by a security advisory. ### How? Assess upgrades once during dev startup and share the result with the CLI nudge. The existing HMR `SYNC` message carries the security advisory to DevTools, which shows a Vulnerability Insight in the existing Insights overlay with a copyable upgrade prompt. The Insight applies to the `security`, `latest`, and `future` policies when the assessment finds an affected version and a safe target. A private test override previews the UI without a vulnerable fixture or live advisory data. ### Screenshots These show the badge and overlay layout. The current copied prompt runs `next upgrade --ai=security` from the app directory. **DevTools badge**  | Light | Dark | | --- | --- | |  |  | <!-- NEXT_JS_LLM --> Closes NAR-939
-
Jiwon Choi authored
Stacked on #99230. ### Why? Simplify the advisory check by querying only the versions we need instead of sending a bulk audit request. ### How? Query npm advisories for the installed version and relevant upgrade candidates. Remove the GHSA lookup and fallback. Only show a nudge when the assessment finds a ready target; otherwise, skip it. <!-- NEXT_JS_LLM --> Closes NAR-948
-
- 26 Sep, 2026 8 commits
-
-
next-js-bot[bot] authored
-
Jiwon Choi authored
Stacked on #99223. ### Why? Stable security advisories can misclassify prerelease versions because prerelease ordering does not reliably establish which security fixes they contain. A blocked upgrade needs a clear way to continue on the installed channel. ### How? Explain the advisory limitation in the security error and give the full `npx next@canary upgrade --ai=latest` command, with instructions to run it from the app directory. Give unsupported Future Defaults requests the same latest-channel recovery command. <!-- NEXT_JS_LLM -->
-
Jiwon Choi authored
Stacked on #99217. ### Why? Apps using RC, beta, or numbered preview releases should be able to upgrade to a newer release on the same prerelease channel. Configured latest reminders should use the same channel selection as explicit upgrades. ### How? Resolve latest targets from the installed channel's npm dist-tag and require the selected target to stay on that channel. Include RC, beta, and numbered preview versions in configured latest reminders. Reject commit-specific preview builds, which have no matching dist-tag, and keep security and unsupported Future Defaults requests blocked. <!-- NEXT_JS_LLM -->
-
Jiwon Choi authored
### Why? Apps without a Git repository should still be able to use agentic upgrades, including upgrades started from a configured reminder. ### How? Check for duplicate work using the repository and provider capabilities available to the app. Direct the agent to upgrade in place when Git is absent, and use Git worktrees, commits, and pull requests when available. <!-- NEXT_JS_LLM -->
-
next-js-bot[bot] authored
-
Luke Sandberg authored
## Summary A dynamically imported local-only ESM module could load through a facade while static named imports resolved to its locals module. For shared singletons, that meant two runtime identities. This backs off the facade split introduced for export mangling in #97770, addressing the remote-components regression reproduced in #99279. Modules with actual re-exports still split as before, and unsplit modules remain eligible for export-name mangling when their names can be changed safely. Escaping namespaces keep their original export names. Focused execution coverage checks that mixed dynamic/static imports register a single module identity; the affected namespace and snapshot expectations are updated. ## Verification - `cargo test -p turbopack-tests --test execution` — 311 passed. - `cargo test -p turbopack-tests --test snapshot` — 132 passed. - Turbopack ECMAScript unit suite — 590 passed. - `pnpm build-all`, `cargo fmt --all -- --check`, Prettier and lint-staged checks passed. - Both #99279 production Turbopack fixtures passed with shared-singleton assertions in a disposable checkout; those unmerged fixtures are not included here. <!-- NEXT_JS_LLM --> <!-- fleet faf1202d-c919-4f7d-bb74-d3fe954bbc57 --> --------- Co-authored-by:
vercel-fleet-prod[bot] <318278635+vercel-fleet-prod[bot]@users.noreply.github.com> Co-authored-by:
Luke Sandberg <210140+lukesandberg@users.noreply.github.com>
-
Janka Uryga authored
we weren't using this flag in any deploy tests so this was fine, but i'm about to
-
Benjamin Woodruff authored
Fixes https://github.com/vercel/next.js/issues/99157 Normally, we only allow crossing roots at symlink boundaries, but the webpack loader could cross roots by following a symlink, and we should allow that.
-