Pull Request Explorer

Summary

Bundle 2.x for n8n

Adds bundled utilities and updates for AI and backend components, improving performance and compatibility.

Health Assessment

• • Fast cycle time and minimal review rounds indicate efficient process; large code changes suggest significant refactoring but no blockers.

Summary

Update bundle 1.x with AI node fixes

This PR updates the 1.x bundle with bug fixes and enhancements for AI parsing and node functionality, improving stability and feature completeness for users.

Health Assessment

• • Rapid merge after cherry‑pick with minimal review indicates high confidence in the changes, but the large code churn warrants careful monitoring for downstream impact.

Summary

Cap auto-detected editor rows within NDV boundary

Prevents SQL query editor from exceeding UI boundaries, ensuring users can view and edit long queries without loss of content. This improves user experience and reduces potential data entry errors.

Health Assessment

• • Fast turnaround with minimal changes and a regression test; review and merge completed within 9 hours, indicating efficient process.

Summary

Add labels to logs view buttons

Improves usability by labeling buttons in the logs view, enabling users to identify actions quickly.

Health Assessment

• • The PR had a long cycle time due to merge conflicts but was reviewed quickly and had minimal friction.

Summary

Add eval cases for evaluation generation

Adds Instance AI workflow evaluation scenarios to enable automated testing of AI agent evaluation creation, improving reliability of AI features.

Health Assessment

• • Fast review and merge within 9 hours, single commit, no rework, indicating smooth process.

Summary

Support host-controlled delegate sub-agent tool surface

Allows hosts to customize the delegate sub-agent tool's name, description, and system instruction, enabling better integration with host systems like Instance AI and preventing duplicate or conflicting text.

Health Assessment

• • Fast review and merge with minimal iterations, indicating clear scope and well-tested changes.

Summary

Fix expression isolate acquisition for node inputs

Ensures dynamic node inputs resolve correctly in VM expression mode, preventing validation errors in the instance‑ai validate‑workflow tool.

Health Assessment

• • Quick fix with minimal changes, fast review, no blockers.

Summary

Fix expression isolate acquisition in node parameters

Ensures AI‑powered node parameter resolution works with the VM expression engine, preventing runtime failures and improving reliability for users relying on AI tools.

Health Assessment

• • Quick, single‑commit fix with minimal code changes and no blockers, resulting in a fast review and merge cycle.

Summary

Hide unsupported n8n Connect actions in node dropdowns

Filters out AI‑gateway‑unsupported actions from node dropdowns, improving UX and preventing errors when using n8n Connect credentials.

Health Assessment

• • Fast review cycle and minimal rework; AI assistance streamlined implementation.

Summary

Auto-name workflow versions via MCP

Adds meaningful names and descriptions to workflow versions created or updated by MCP clients, improving traceability and user experience.

Health Assessment

• • The PR introduced extensive changes across many files, but was reviewed quickly. Multiple AI-assisted reviews helped identify issues, leading to a final approval after several iterations.

Summary

Add Instance AI eval cases for builder changes

Adds regression guard test cases for recent builder changes, ensuring new behavior is validated and preventing future failures.

Health Assessment

• • Rapid review and merge with minimal changes indicates low complexity and high confidence in test additions.

Summary

Fix vector store provider error handling

Improves user experience by converting raw provider errors into clear node errors, reducing confusion and Sentry events.

Health Assessment

• • The PR resolved a critical error handling issue, received a quick initial review, but required several follow‑up commits to refine error normalization and preserve provider messages.

Summary

Fix concurrent executions label when manual run is active

Fixes a visual bug where the concurrent executions header incorrectly shows 'No active executions' during a manual workflow run, improving UI accuracy and user experience.

Health Assessment

• • The PR was reviewed quickly, with a single comment and a subsequent approval after a brief rework, indicating efficient collaboration.

Summary

Add workflow reviews toggle to Security & policies

Adds an instance‑level toggle for workflow reviews, gated by license and feature flag, enhancing security control and user flexibility.

Health Assessment

• • Review was rapid and issue‑free, but four commits followed the initial review, indicating iterative refinement. The PR added 587 lines across 20 files, a sizable change that still maintained a fast overall cycle time.

Summary

Add Pinecone vector store backend to agents SDK

Adds Pinecone vector store support, expanding vector store options for agents and enabling scalable vector search integration.

Health Assessment

• • Merged within 0.6 hours with minimal review, indicating high confidence in code quality and low risk.

Summary

Add hover state to N8nSwitch component

Adds a subtle hover effect to the switch component, improving perceived responsiveness and usability for users.

Health Assessment

• • Quick review with minimal rework, small scope, fast cycle time.

Summary

Decouple DB from blob store interface

This refactor separates the database logic from the blob store interface, enhancing modularity and simplifying future storage integrations.

Health Assessment

• • The PR underwent a single AI-assisted review cycle, taking over four days to merge, indicating moderate review friction for a straightforward refactor.

Summary

Follow redirects when loading Keycloak authorization page

Fixes intermittent 302 responses from Keycloak by following redirects and accumulating cookies, restoring e2e tests and improving authentication reliability.

Health Assessment

• • Quick, single-file change with minimal review and no rework; AI-generated summary streamlined communication.

Summary

Extend Instance AI intent eval fixtures

Broadens Instance AI intent evaluation coverage across workflows, hybrids, agents, and ambiguous requests, adding new test cases and shared schemas to improve classification accuracy.

Health Assessment

• • The PR underwent a lengthy review process with multiple merge conflicts and extensive test additions, indicating significant rework and potential integration complexity.

Summary

Add breaking change rules for v3

Adds comprehensive breaking‑change detection for v3 migrations, ensuring migration reports cover all deprecation warnings.

Health Assessment

• • Rapid review and merge with minimal friction, but large code addition introduces moderate risk.

Summary

Backport: Postgres-only DB connection recovery

Ensures database recovery logic applies only to Postgres, preventing crashes in SQLite environments under load.

Health Assessment

• • Quick backport merged within minutes, indicating low complexity and high confidence in the fix.

Summary

Expose a composable Scheduler surface from @n8n/scheduler

Refactors the scheduler into pure logic, removing dependency injection and database coupling, enabling the CLI to implement a DurableScheduler without changing behavior and laying groundwork for future scheduler features.

Health Assessment

• • Large refactor completed in under a day with minimal review, indicating high confidence in code quality and thorough testing.

Summary

Fix database connection recovery for Postgres only

Ensures that the database connection recovery logic is executed only for Postgres, preventing crashes on SQLite by avoiding unnecessary pool teardown and improving stability under load.

Health Assessment

• • Quick turnaround with minimal rework; single review cycle and small scope indicate low risk.

Summary

Fix leaked assistant timer on stream stop

Prevents flaky test failures and ensures reliable assistant functionality, improving developer experience and CI stability.

Health Assessment

• • Long cycle time indicates delayed review, but minimal changes and a single commit suggest low complexity and low risk.

Summary

Fix API playground URL slash join

Corrects malformed API playground links, ensuring users reach the correct Swagger UI or docs page.

Health Assessment

• • The PR was merged after a long review cycle, indicating a backlog or scheduling delay, but the change itself was small and low risk.

Summary

Stop executions list from firing endless load-more requests

Fixes infinite load-more requests in executions list, preventing unnecessary server calls and improving UI responsiveness for users.

Health Assessment

• • The PR had a slow review cycle but minimal rework, indicating a straightforward bug fix with moderate review time.

Summary

Export projects with workflows and folders

Adds the ability to export entire n8n projects—including workflows, folders, and credentials—into a package for sharing or backup, enhancing portability and backup capabilities.

Health Assessment

• • The PR required multiple commits after a single review, indicating moderate review friction. The large code change (≈474 lines) and a 4‑day cycle time suggest a higher risk of integration issues, but the single review round mitigates some risk.

Summary

Fix AWS request signing to include query parameters

Ensures query parameters entered in the HTTP Request node are correctly signed and transmitted to AWS services, preventing silent data loss and improving API reliability.

Health Assessment

• • The PR was resolved in a single commit with a quick review cycle, indicating minimal friction and low risk to the codebase.

Summary

Add Supabase vector store backend to agents SDK

Adds SupabaseVectorStore backend enabling agents to use Supabase for vector storage, expanding supported vector stores and improving integration flexibility.

Health Assessment

• • Rapid 1.8‑hour cycle with a single review round indicates low friction and high confidence in the changes.

Summary

Add Qdrant vector store backend to agents SDK

Adds a Qdrant backend for the agents SDK, enabling vector storage via Qdrant and improving scalability and flexibility for vector operations.

Health Assessment

• • Fast cycle time and minimal review comments indicate smooth integration.

Summary

Allow custom OAuth2 scopes for Zendesk

Enables users to request additional Zendesk permissions beyond the default read/write, improving integration flexibility and user control.

Health Assessment

• • Fast review and merge with minimal rework; clear test coverage and straightforward implementation.

Summary

Fix AI Agent Node tool call redundancy

Reduces LLM confusion and token usage by removing redundant tool call messages and clarifying chat tool responses, improving reliability and cost efficiency.

Health Assessment

• • Quick review and merge within a day, minimal rework, indicates straightforward bug fix.

Summary

Refactor expression isolate acquisition in webhook registrar

Improves isolation handling for webhook triggers, ensuring safe concurrent execution and preventing runtime errors.

Health Assessment

• • The PR refactors expression isolation logic with a single review after a long review cycle, indicating moderate complexity but low risk of breaking functionality.

Summary

Persist MCP Server Trigger tool calls in queue mode

Tool calls executed through MCP Server Trigger are now correctly persisted in execution history. This improves traceability and debugging for users.

Health Assessment

• • Fast review with minimal back‑and‑forth, but the change adds over 500 lines of code, indicating a substantial refactor.

Summary

Fix Postgres pool teardown timeout during recovery

Prevents n8n from hanging and serving 503 errors when Postgres connections become unreachable, improving uptime and reliability for users.

Health Assessment

• • The PR took over 10 days from opening to merge with multiple commits and a long review time, indicating significant complexity and potential risk, but ultimately resolved a critical reliability issue.

Summary

Fix stale i18n assertion in test

Updates test to use dynamic i18n lookup, preventing false failures when UI text changes. Maintains test stability across future copy updates.

Health Assessment

• • Quick fix with minimal changes, fast review, no blockers.

Summary

Support custom global roles in token exchange

Enables external systems to map custom roles to n8n users during token exchange, expanding integration flexibility and ensuring proper licensing enforcement.

Health Assessment

• • The PR was merged within 15 minutes of opening, with no review comments or rework, indicating a straightforward, low-risk change.

Summary

Surface red scenarios in AI eval skill

Improves documentation to highlight real build issues rather than workarounds, ensuring accurate calibration of AI workflow evaluation.

Health Assessment

• • Docs-only change with quick review and minimal edits, indicating low risk and high confidence in the update.

Summary

Add custom global role support in token exchange

Enables external systems to assign custom global roles during token exchange, expanding role flexibility and integration capabilities.

Health Assessment

• • Fast cycle time and minimal review comments indicate a smooth, low-risk change; AI-generated description helped streamline communication.

Summary

Remove Instance AI delegate orchestration tool

Simplifies orchestration by removing unused delegate tool, reducing complexity and maintenance overhead.

Health Assessment

• • The PR involved extensive refactoring across many files, with rapid initial review but subsequent iterative commits, indicating moderate review friction and a high‑risk, large‑scope change.

Summary

Make eval scenario state deterministic

Fixes deterministic state for eval scenarios by preserving empty pins and clearing static data, ensuring consistent behavior and preventing data leakage.

Health Assessment

• • The PR required over 82 hours to merge with a single review comment, indicating a slow review cycle but minimal rework. The changes span 227 lines across four files, a large scope that necessitates thorough testing to ensure deterministic behavior.

Summary

Improve AI verifier resilience and scoring

Enhances reliability of AI evaluation by preventing verifier stalls and correctly excluding incomplete runs from failure metrics, reducing false positives and improving accuracy of evaluation results.

Health Assessment

• • The PR required a lengthy review period (~57h) but had no subsequent rework, indicating a thorough initial submission. The large code changes (~1k lines) suggest significant impact on evaluation logic, but the single review round and no rework reduce risk.

Summary

Report node errors in workflow verification

Adds node-level error reporting to workflow verification, ensuring that workflows with continue‑on‑fail nodes are not incorrectly marked as fully verified.

Health Assessment

• • Fast review and quick iteration; single review round; AI bot review helped catch issues.

Summary

Guard against null run-data slots in execution redaction

Fixes a critical runtime error that caused execution list requests to fail when encountering null entries in run data, preventing 500 errors and improving reliability for affected users.

Health Assessment

• • Rapid resolution with minimal changes and no review needed, indicating a straightforward bug fix.

Summary

Guard against null run-data slots in execution redaction

Fixes a crash caused by null entries in execution data, preventing 500 errors for users. Improves reliability of execution list views and reduces support incidents.

Health Assessment

• • Backport merged within 14 minutes, indicating straightforward fix and minimal review friction.

Summary

De-flake node creator search navigation test

Fixes flaky e2e test by asserting node visibility instead of count, improving reliability of node search functionality.

Health Assessment

• • Quick fix with minimal changes, resolved within a day, indicating efficient review and low complexity.

Summary

Fix client confirmation mode for non-cloud n8n

Ensures computer‑use connections to non‑cloud instances use client confirmation mode, improving security and user experience.

Health Assessment

• • Quick review and single commit indicate a straightforward fix; the long cycle time reflects scheduling rather than complexity.

Summary

Rename private credentials to end-user credentials

Unifies terminology across the platform, improving user experience and reducing confusion around credential types.

Health Assessment

• • Fast review and minimal rework despite a large code change; the rename is straightforward but touches many files, indicating moderate risk but low friction in the review process.

Summary

Guard against null run-data slots in execution redaction

Fixes a runtime crash in execution redaction by adding null checks, preventing 500 errors for users and ensuring dynamic credential redaction continues to work.

Health Assessment

• • Quick fix with minimal changes, approved within 1.5 hours, indicating low complexity and high confidence.

Summary

Fix credential connections on private credential move

Reconcile per-user connections when a private credential is moved to another project, ensuring access is properly updated.

Health Assessment

• • The PR was reviewed and approved quickly, indicating a smooth review process.
• • The changes are moderately sized, but well-structured and easy to follow.