Pull Request Explorer

Summary

Send User-Agent and x-client-app headers in harness adapters

Adds versioned client attribution headers to harness adapters, ensuring consistent client identification across AI Gateway requests.

Health Assessment

• • Fast cycle time (1.8h) and minimal review friction (0.2h) indicate a straightforward, well-scoped change with low risk.

Summary

Promote throttle option to stable in React SDK

Enables developers to use a stable throttle setting for chat and completion hooks, improving performance and reducing unnecessary API calls.

Health Assessment

• • PR was reviewed and merged within 48 minutes, single commit, minimal code churn, indicating a low‑risk, high‑confidence change.

Summary

Add functions config support to @vercel/container

Enables container services to honor vercel.json functions settings for memory, maxDuration, regions, and queue triggers, improving resource allocation and deployment consistency.

Health Assessment

• • Fast review (0.1h) and only two follow‑up commits indicate low friction. The feature adds moderate code changes (~200 lines) and is supported by AI assistance, resulting in a straightforward, low‑risk merge.

Summary

Split large cache-components-dev-warmup test suite

Splits slow cache-components-dev-warmup tests into smaller suites, enabling parallel execution and faster CI feedback.

Health Assessment

• • Fast turnaround with minimal changes, indicating straightforward test refactor.

Summary

Fix missing step numbers in harness telemetry

Ensures telemetry events include step numbers, preventing incomplete run states and improving reliability of harness monitoring.

Health Assessment

• • Quick fix with minimal changes and fast review cycle indicates low complexity and low risk.

Summary

Cancel superseded Server Components HMR refresh

Improves development experience by aborting server-side work for superseded HMR refreshes, reducing unnecessary computations and validation.

Health Assessment

• • The PR has a moderate review time and a reasonable number of commits, indicating a manageable development process.

Summary

Add documentation for prepareSandboxForHarness

Adds a docblock for the prepareSandboxForHarness utility and renames a helper file to match its function name, improving code clarity and maintainability.

Health Assessment

• • Quick turnaround with minimal changes and fast review indicates low complexity and high confidence in documentation update.

Summary

Backport Gemini Interactions API to v6

Adds Gemini Interactions API support to @ai-sdk/google-vertex, enabling new interaction capabilities for users.

Health Assessment

• • Quick turnaround with a single commit and minimal review rounds indicates a straightforward feature addition.

Summary

Backport: Gemini Interactions video output support

Adds video output parsing and per‑modality token breakdown for Gemini Interactions, enabling richer media handling and analytics.

Health Assessment

• • Quick turnaround with a single commit and minimal review, indicating a straightforward implementation.

Summary

Add sandbox abstraction architecture doc

Provides a new architecture document explaining sandbox abstraction and its interaction with harnesses, improving developer clarity and onboarding.

Health Assessment

• • Rapid merge with minimal changes and quick review indicates low complexity and high confidence in documentation quality.

Summary

Pre-approve esbuild build script in pnpm upgrades

Ensures Vercel CLI upgrades via pnpm run without interactive prompts, improving automation and reliability for developers.

Health Assessment

• • Quick iteration to add --allow-build flag, minimal code changes, fast merge, low risk.

Summary

Fix CLI self-upgrade for pnpm 11 installs

Fixes mis-detection of pnpm 11 global installs, preventing accidental local installs during CLI self-upgrade. Also removes redundant upgrade prompts after successful upgrades, improving user experience.

Health Assessment

• • The PR required multiple commits after the first review, indicating moderate review friction. The changes were limited to CLI logic and tests, keeping scope medium and risk moderate.

Summary

Update gateway model settings files

Automated update of gateway model settings to surface newly released models as IDE autocomplete suggestions.

Health Assessment

• • Fast automated merge with minimal changes, no review needed.

Summary

Backport: update gateway model settings

Automated update of gateway model settings to surface new models in IDE autocomplete.

Health Assessment

• • Quick automated backport with minimal changes, merged within minutes.

Summary

Enable bytecode caching for lambda functions

Adds bytecode caching to reduce deployment size and improve cold start times for Python lambda functions, ensuring builds stay within size limits.

Health Assessment

• • Long review cycle and multiple commits indicate significant rework; large code changes may increase risk of build failures.

Summary

Fix immutable static asset deployment tests

Ensures deployments using immutable static assets correctly unset the asset token and that search parameters return undefined for unknown keys, improving test reliability.

Health Assessment

• • Quick turnaround with minimal changes indicates a straightforward fix with low risk.

Summary

Migrate webpack tests back to arm runners

Restores ARM-based runners for webpack tests, reducing CI queue times and improving build reliability.

Health Assessment

• • Fast review and merge with minimal changes to CI pipeline, indicating low risk and high confidence.

Summary

Fix testing-adapters variable name

Corrects variable name in testing adapters documentation, improving clarity for developers.

Health Assessment

• • Quick documentation fix merged within 30 minutes, minimal review needed.

Summary

Use Turbopack for test deployment

Adds Turbopack support for running tests on deployment, improving test speed and reliability.

Health Assessment

• • The PR was reviewed and merged within 7.3 hours, indicating a smooth process with minimal friction. The changes are small to medium in scope and pose low risk to the codebase.

Summary

Disable failing service worker deploy tests

Temporarily disables failing service worker deployment tests to keep CI stable while investigation continues.

Health Assessment

• • Quick resolution with minimal changes, no rework required.

Summary

Fix next info analyze graph css chunking

This change ensures Next.js analyze correctly sets the TURBOPACK environment flag, preventing build failures. It also skips CSS chunking validation during the next info phase, improving stability for developers.

Health Assessment

• • Fast review and minimal changes indicate low complexity and quick resolution.

Summary

Bump graphite-ci-action in CI workflow

Updates the CI workflow to use the latest graphite-ci-action, improving build reliability.

Health Assessment

• • Quick merge with minimal changes indicates low complexity and high confidence in the update.

Summary

Fix HTTP MCP mid-stream disconnect crash

Prevents server crashes caused by mid-stream disconnects in HTTP MCP, improving reliability for downstream services.

Health Assessment

• • Merged within 12 minutes, indicating a straightforward bug fix with minimal impact.

Summary

Prevent HTTP MCP mid-stream disconnects from crashing process

Fixes unhandled promise rejections caused by mid-stream server disconnects in HTTP MCP, ensuring Node processes remain stable.

Health Assessment

• • Quick fix with minimal changes, merged within minutes, indicating low complexity and high confidence.

Summary

Prevent HTTP MCP mid-stream disconnect crashes

This fix stops unexpected process crashes caused by mid-stream disconnects in HTTP MCP, improving system reliability and uptime.

Health Assessment

• • The PR resolved a critical crash issue with minimal changes and a quick turnaround, indicating efficient issue triage and robust testing.

Summary

Add instrumentation for Node.js middleware without adapter

Ensures instrumentation works for legacy Node.js middleware setups, improving observability and debugging for developers.

Health Assessment

• • Quick fix with minimal changes, resolved within a single review cycle.

Summary

Fix Mistral presencePenalty and frequencyPenalty reporting

Ensures Mistral chat requests correctly forward presence and frequency penalties, eliminating unsupported warnings and improving model behavior.

Health Assessment

• • Quick fix with minimal changes, merged within minutes, indicating low complexity and high confidence.

Summary

Fix Mistral penalty reporting bug

Corrects Mistral chat request builder to forward presence and frequency penalties, eliminating unsupported warnings and ensuring accurate request payloads.

Health Assessment

• • Rapid backport and minimal changes indicate a straightforward bug fix with low complexity.

Summary

Fix Mistral penalty reporting bug

Corrects Mistral chat request builder to forward presence and frequency penalties, eliminating unsupported warnings and ensuring accurate request payloads.

Health Assessment

• • Quick fix merged within 30 minutes of opening, minimal code changes, indicating low complexity and high confidence.

Summary

Version Packages

This PR updates package versions and changelogs to prepare for a new release, enabling automated npm publishing.

Health Assessment

• • Fast merge with minimal changes, typical release automation.

Summary

Propagate maxDuration to Edge adapter outputs

Improve deployment adapters by propagating maxDuration metadata for consistent route execution limits across Node.js and Edge outputs.

Health Assessment

• • The PR had a relatively long cycle time of 310.2 hours, but the review process was straightforward with minimal friction.

Summary

Skip apt-installed dependencies in Playwright install

Reduces test setup time by ~15s per Chromium job, saving ~$0.47 per PR push and ~$1,800/month in runner costs.

Health Assessment

• • Fast review and minimal changes led to quick merge, indicating low complexity.

Summary

Fix service worker error test

Corrects a failing service worker deployment test, improving reliability of e2e testing.

Health Assessment

• • Fast cycle time with a single approval and no rework, indicating a low-risk, straightforward fix.

Summary

Upgrade SWC to v72 and Rust React Compiler

This upgrade improves build performance and compatibility for Next.js applications by updating the SWC compiler and Rust React compiler, ensuring faster compilation and better support for modern JavaScript features.

Health Assessment

• • Fast cycle time and minimal review rounds indicate smooth integration; moderate code changes were quickly approved, reflecting low complexity and high confidence in the upgrade.

Summary

Update gh-stack skill

Fixes gh-stack skill to remove deprecated --draft option, aligning with upstream changes.

Health Assessment

• • Quick turnaround with minimal rework; PR resolved within a few hours and merged promptly.

Summary

Update Cache Components Caching Documentation

Enhances Next.js documentation to clarify caching mechanisms, improving developer understanding and reducing support queries.

Health Assessment

• • The PR had a long cycle time due to delayed review, but the changes were straightforward and low risk.

Summary

Add --project flag support for CLI commands

Enables CLI commands to run from any directory by specifying a project, improving CI and agent reliability and reducing failure rates.

Health Assessment

• • The PR required extensive rework over a week, indicating complex changes and potential integration challenges.

Summary

Fix framework detection for subdirectory entrypoints

The change ensures framework detection runs at the service root, not the entrypoint directory, preventing silent failures and incorrect framework assignments for builders.

Health Assessment

• • Long review time indicates delayed feedback; the PR was merged after a week, but the issue was straightforward and resolved quickly.

Summary

Expose totalUsage and finishReason on stream result

Adds missing usage metrics and finish reason to workflow stream results, aligning API consistency. This improves developer experience and debugging.

Health Assessment

• • Quick fix with minimal changes, fast review and merge, indicating low complexity and high confidence.

Summary

Reduce payload in doStreamStep for performance

This change trims unnecessary data from the workflow event log, improving performance and reducing storage costs while preserving the user‑facing StepResult shape. The optimization reduces payload size, speeding up processing and lowering storage usage.

Health Assessment

• • Long cycle time and single review after a month indicates low review activity; the change was a large refactor that required a single approval after a long wait.

Summary

Remove incorrect force-cache default statement

Clarifies fetch caching behavior for Next.js 15, preventing developer confusion and potential performance issues.

Health Assessment

• • The PR had a long cycle time and multiple commits, indicating extended review and rework, but the change is small and low risk.

Summary

Preserve generated service routes

Ensures that generated service routes are retained during CLI build output assembly, preventing missing root service handoff routes when using custom output directories.

Health Assessment

• • Quick turnaround with minimal rework; PR addressed a specific bug and merged within 2 hours.

Summary

Version Packages

Automated release of Vercel packages, updating changelogs and package versions for consistent publishing.

Health Assessment

• • Release PR merged quickly with minimal changes, indicating an efficient and automated release workflow.

Summary

Fix Vercel CLI skill documentation

Improves agent guidance for Vercel CLI usage, reducing failure rates by clarifying flags, output handling, and pagination.

Health Assessment

• • The PR required multiple iterations over documentation to align with production failure analysis, but no code changes were made, keeping risk low.

Summary

Add project set command to CLI

Enables users to update project presets directly from the CLI, improving workflow and reducing reliance on the dashboard.

Health Assessment

• • The PR required over 10 days to merge, with multiple commits after the initial review, indicating moderate complexity and a higher risk of impacting deployments.

Summary

Buffer tool call deltas until function name arrives

Fixes error when OpenAI-compatible providers send tool-call deltas without function name, ensuring proper buffering and error handling. Improves reliability of streaming responses across providers.

Health Assessment

• • Single commit, quick merge, minimal review friction, small number of files, moderate line changes.

Summary

Fix stale transport in useChat hook

Ensures the useChat hook always uses the most recent transport, preventing stale connections and improving reliability for real‑time chat features.

Health Assessment

• • Quick fix with minimal changes, merged within an hour, indicating low complexity and high confidence in the change.

Summary

Fix default tool execution denial message

Improves tool call handling for GPT-5 models by updating the default denial message, preventing premature rejection of subsequent calls. This ensures smoother integration for developers using the AI SDK.

Health Assessment

• • Quick fix merged within an hour, indicating minimal review friction and low risk.

Summary

Fix tool input missing in dynamic tool results

Ensures provider‑executed dynamic tools correctly record input, improving reliability of tool result streams.

Health Assessment

• • Rapid resolution within an hour, minimal changes, low complexity and risk.

Summary

Fix step/chunk timeout aborts with TimeoutError reason

Ensures timeouts in text generation are clearly identified, improving error handling for consumers.

Health Assessment

• • Quick, single-commit fix with minimal scope and no review back‑and‑forth.