Pull Request Explorer

Summary

Fix InputTriple responsive behaviour in NDV parameters

Restores responsive layout of the InputTriple component in the NDV parameters panel, improving UI consistency and preventing layout regressions.

Health Assessment

• • Rapid backport with minimal changes, fast review and merge, low risk.

Summary

Fix custom header credential inconsistencies

Ensures model list loading works even when custom header expressions fail to evaluate, improving reliability of OpenAI and Anthropic integrations.

Health Assessment

• • Quick resolution with minimal rework; single review cycle and fast response time.

Summary

Fix eval build traces to separate LangSmith project

By directing eval build traces to a dedicated LangSmith project, the PR prevents contamination of real-user telemetry data. This ensures more accurate trace analysis and reliable metrics for future evaluations.

Health Assessment

• • Quick review and merge with minimal changes; AI review bot flagged no issues.

Summary

Record git provenance metadata for eval experiments

Adds CI metadata capture for eval experiments, enabling branch-based KPI tracking and distinguishing local vs CI runs. This improves analytics accuracy and supports dev‑vs‑official split.

Health Assessment

• • Fast review but long cycle due to merge timing; minimal rework; small scope.

Summary

Fix 5 security issues via dependency updates

Updated hono, vue-i18n, and linkify-it to address DoS, XSS, and header parsing vulnerabilities, eliminating multiple high‑severity CVEs.

Health Assessment

• • Automated dependency update resolved multiple CVEs with minimal code changes, demonstrating efficient security patching.

Summary

Improve Windows build resilience with retry

Adds retry logic and concurrency cap to the Windows CI job, reducing flaky failures and ensuring reliable builds for developers.

Health Assessment

• • Fast cycle time (1.6h) and a single review comment indicate minimal friction and low risk.

Summary

Drop DB-rooted Sentry transactions

Reduces tracing noise by removing root DB transaction spans, improving observability clarity and cutting log volume.

Health Assessment

• • Quick approval with minimal changes, indicating straightforward maintenance.

Summary

Add feature flag for separate browser use

Users can toggle separate browser use in AI Assistant settings, enabling or disabling browser functionality via a feature flag, improving flexibility and reducing confusion.

Health Assessment

• • The PR introduced a new feature flag to control browser use in AI Assistant, with a quick initial review but required several small commits to resolve merge conflicts and tests, indicating moderate iteration but overall smooth process.

Summary

Fail loud on unmatched chat-hub proxy requests

Adds a guard to make unmatched proxy requests fail loudly, surfacing drifted request bodies for clearer debugging.

Health Assessment

• • Quick review and single commit indicate a smooth process; the change is a small test‑infrastructure tweak with minimal risk to production.

Summary

Serialize OAuth2 credential refresh to prevent token invalidation

Prevents users from having to reconnect when concurrent refreshes invalidate rotating tokens, improving reliability and reducing support overhead.

Health Assessment

• • The PR required multiple commits after review and introduced a new lock service, indicating moderate complexity but was resolved quickly with a fast review turnaround.

Summary

Show alert when selected time range for insights lacks data

Adds a user‑facing alert in the insights dashboard to inform users when the selected time range contains no data, improving clarity and reducing confusion.

Health Assessment

• • Fast cycle time (11.4h) and near‑instant review (0.1h) indicate a smooth process. The feature is well scoped with minimal rework, and the change is low risk.

Summary

Refactor: Consolidate PEM formatting into utils

Centralizes PEM block formatting logic, reducing duplication and simplifying imports across the n8n codebase. This change improves maintainability and prevents future import cycles.

Health Assessment

• • The PR involved a substantial refactor across many files but was reviewed quickly with minimal friction, indicating a well‑structured change.

Summary

Fix Anthropic reasoning preservation through proxy

Ensures AI Assistant reasoning is correctly handled when using proxied Anthropic models, preventing noisy warnings and maintaining valid reasoning.

Health Assessment

• • Quick turnaround with minimal rework; single review comment resolved swiftly.

Summary

Prevent builder from re-running verified workflow

Ensures the AI assistant builder does not trigger a live run after verification, eliminating unnecessary approval prompts and improving user experience.

Health Assessment

• • Quick review and single commit indicate a straightforward bug fix with minimal risk.

Summary

Fix WhatsApp Trigger verify-token guidance

Corrects AI hallucination about WhatsApp verify token, adding regression tests to ensure accurate guidance and prevent misconfiguration errors.

Health Assessment

• • Review took long (69.5h) but no rework was needed, indicating a straightforward fix.

Summary

Add caching for published workflow version lookup

Improves performance by caching published workflow data, reducing database load on trigger execution and ensuring consistency during republishing.

Health Assessment

• • Multiple AI‑assisted commits and extensive review comments led to a 10‑day cycle; the caching feature reduces DB load but requires careful invalidation logic.

Summary

Release 2.28.3

Updates core to prevent startup failures from partially-installed community packages and bumps package versions for a stable release.

Health Assessment

• • Merged in 18 minutes with no review friction and minimal code changes.

Summary

Release 2.27.5

Patch release that includes a bug fix preventing startup failure from partially-installed community packages.

Health Assessment

• • Rapid release with minimal changes and no review required.

Summary

Align MCP OAuth e2e with decoupled access setting

Updates the end‑to‑end test to reflect the new OAuth behavior when MCP access is disabled, ensuring tests pass and match the production design.

Health Assessment

• • Test updated to match new OAuth behavior; minimal changes, quick review, low risk.

Summary

Reduce Daytona sandbox auto‑archive and delete defaults

Lower idle window for Daytona sandboxes, freeing storage faster and reducing long‑term accumulation.

Health Assessment

• • Fast review (0.9h) and minimal changes (10 lines) indicate low risk and efficient process.

Summary

Fix import of minimist from zx in benchmark script

Corrects a runtime error in nightly benchmark runs by importing minimist from the declared zx dependency, ensuring consistent script behavior without adding new dependencies.

Health Assessment

• • Quick fix with minimal changes, resolved within minutes, indicating efficient review and low risk.

Summary

Add PR-aware dispatch for Instance AI evals

Enables re-running AI evaluation workflows against the current PR head, improving test coverage and reducing stale results.

Health Assessment

• • The PR added a new dispatch path for AI evals with minimal rework, received a single review, and merged within two days, indicating a smooth process.

Summary

Sanitize workflow inputs for CI pipelines

Ensures CI workflows use sanitized inputs, improving security and reliability.

Health Assessment

• • Quick review and minimal changes indicate a low‑risk, low‑friction PR.

Summary

Add credential:connect scope for private credentials

Enables users with workflow execution rights to run OAuth authorization on private credentials without needing edit permissions, streamlining automation while maintaining security controls.

Health Assessment

• • The PR received a rapid review and was merged after a single iteration, indicating a smooth process. The moderate code churn (≈150 lines across 16 files) reflects a focused feature addition. The overall cycle time was longer than ideal, likely due to merge timing rather than technical complexity.

Summary

Update node popularity data

Automated weekly update of node popularity data to reflect current usage patterns, improving node sorting in the editor UI.

Health Assessment

• • Automated PR with no review needed, merged within 0.3 hours, minimal risk.