Pull Request Explorer

Summary

Enable multiturn agent building in instance AI

Adds multi‑turn conversation support for agent building, allowing the assistant to edit the same agent across turns and enforce project scoping.

Health Assessment

• • Fast cycle time and quick review indicate efficient collaboration; the large code change and multiple review rounds suggest moderate complexity.

Summary

Route selector‑purity locators through page objects

Refactors Playwright tests to move chained selectors into page objects, clearing selector‑purity violations and refreshing the baseline.

Health Assessment

• • Fast cycle time (0.8h) and minimal review friction indicate a straightforward refactor with low risk. The AI‑assisted changes were reviewed and approved quickly, and the scope of 125 lines across 8 files is moderate.

Summary

Reorder credential type cards to show Fixed first

This change improves user experience by placing the default Fixed credential type card on the left, making it more intuitive for users to select the default option.

Health Assessment

• • Quick review and merge with minimal changes indicates low complexity and high confidence.

Summary

Add private credential usage telemetry

Adds telemetry to track private credential usage and resolution in workflows, enabling insights into adoption and success rates to improve user experience and compliance.

Health Assessment

• • Quick review and single commit indicate a smooth process; the telemetry addition provides valuable analytics with minimal risk.

Summary

Stabilize NDV run selector interactions in vector stores e2e test

Fixes flaky end‑to‑end tests for NDV run selector interactions, improving reliability of vector store workflows.

Health Assessment

• • Quick fix with minimal changes, fast review and merge, low risk.

Summary

Reduce memory usage in source control status

Prevents out‑of‑memory crashes and improves performance for large workflow instances, ensuring reliable source control operations.

Health Assessment

• • Fast review cycle with minimal rework, indicating a straightforward performance improvement.

Summary

Fix webhook auth for editor tests and scope realm

This change stops basic auth dialogs during editor test runs, restoring expected behavior. It also scopes authentication realms per webhook, preventing credential leakage between chats.

Health Assessment

• • Fast turnaround with minimal rework; single review comment indicates straightforward fix.

Summary

Add plugin prefix support for MCP skillsUsed

Enables MCP clients to report plugin‑prefixed skill identifiers, improving telemetry accuracy and distinguishing official from community skills.

Health Assessment

• • Quick approval with minimal rework; AI‑assisted code generation streamlined changes.

Summary

Add AI Assistant template examples experiment

Enables AI-generated workflow templates, enhancing user productivity by providing prompt examples.

Health Assessment

• • Rapid merge after backport with minimal review, but the addition of ~280k lines across 20 files indicates a substantial feature addition that may introduce integration complexity.

Summary

Update Supabase credential description to new secret keys

Updates Supabase credential UI to reflect new secret key naming and tooltips, ensuring compatibility with upcoming key deprecation. No functional changes, but improves user experience and documentation.

Health Assessment

• • Fast review and merge with minimal changes; no rework required.

Summary

Fix auto-bound credential reporting in AI workflow builds

Ensures users are not prompted to reconnect credentials that are already bound, improving workflow creation experience and reducing friction.

Health Assessment

• • Fast cycle time (5.6h) and single review round indicate a straightforward bug fix with minimal rework. AI review identified a single issue, which was resolved quickly.

Summary

Fix stale AI simulation plan before verification

Ensures workflow re‑verification uses updated credentials, preventing false credential errors and improving reliability for users.

Health Assessment

• • Rapid resolution with a single commit and quick review indicates a straightforward bug fix with minimal risk.

Summary

Route NDV selector violations through page objects

Improves test reliability by routing NDV tests through proper page objects, eliminating selector‑purity errors and ensuring accurate UI interactions.

Health Assessment

• • The PR had a long cycle time and a single review round, indicating low urgency but significant test refactoring effort. The changes are non‑critical to business features but improve test stability.

Summary

Add durable scheduler schedule-trigger task handler

Introduces a handler that turns due schedule-trigger events into workflow executions, ensuring exactly‑once processing via deduplication.

Health Assessment

• • Fast AI‑assisted review with minimal back‑and‑forth, but the addition spans over 500 lines across seven files, indicating a substantial backend change.

Summary

Support custom OpenAI-compatible model endpoints in AI Assistant

Adds support for custom OpenAI‑compatible model endpoints, enabling the AI Assistant to connect to local or third‑party LLM servers such as LM Studio, vLLM, or Ollama, thereby improving flexibility and reducing vendor lock‑in.

Health Assessment

• • Fast review and merge with minimal changes, indicating low complexity and risk.

Summary

Port agent builder skills to Instance AI

Adds agent‑builder capability to Instance AI, enabling creation and configuration of n8n agents via a new skill and tool, expanding AI assistant functionality.

Health Assessment

• • Long cycle time with multiple review rounds and large scope indicates significant rework and potential risk. AI review tool cubic flagged issues early, but many changes were required to address them.

Summary

Use correct Gmail OAuth2 credential type in eval credential seeder

Fixes credential type mismatch causing eval cases to fail, ensuring Gmail credentials are correctly registered and workflows run successfully.

Health Assessment

• • Quick fix with minimal changes, reviewed and merged within hours, indicating low complexity and risk.

Summary

Add AI activity step to design system

Introduces reusable AI activity components to the design system, enabling richer tool call displays and improved UI in the editor for a better user experience.

Health Assessment

• • Large scope with 20 files and 1,324 lines added/removed required multiple commits after a single review, indicating significant rework; review was quick but the complexity and size raise risk for regressions.

Summary

Add telemetry event for stalled builder generation

Introduces a telemetry event that triggers when builder generation stalls, enabling better monitoring and reducing user frustration.

Health Assessment

• • Fast cycle time of 1.3h and a single review with no rework indicate a smooth, low-risk PR process.

Summary

Unify onboarding empty states across tabs

Standardizes empty state UI across overview and project tabs, improving consistency and reducing user confusion. Also fixes flash and flicker issues for smoother navigation.

Health Assessment

• • The PR had a moderate cycle time with a single review round, but the large scope and slow initial review suggest moderate risk.

Summary

Fix credential connect banner positioning

Ensures the connect/connected banner remains below the type selector, preventing UI jumps when toggling credential types.

Health Assessment

• • Quick AI-assisted fix with minimal changes, single review, merged within 4 hours.

Summary

Add AI Assistant template examples experiment

Provides AI-generated workflow templates to streamline user onboarding and reduce setup time.

Health Assessment

• • Merged within 0.4 hours after a single review, indicating a smooth approval process. The PR adds a large data payload (~278k lines) but minimal code changes, suggesting low functional risk.

Summary

Add AI Assistant template examples experiment

Adds a catalog of workflow template examples for the AI Assistant, improving user onboarding and reducing setup time.

Health Assessment

• • Large-scale UI and data changes with many commits but quick review indicates confidence; however, the massive data file and numerous changes raise risk.

Summary

Run durable scheduler lifecycle loops on every main

Adds jittered lifecycle loop to the scheduler, enabling concurrent main instances to share work without leader gating, improving reliability and scalability.

Health Assessment

• • Fast cycle time and minimal review comments indicate smooth integration, though the change adds substantial new code.

Summary

Add tooltips to node hover actions

Adds consistent, styled tooltips to node hover actions on the workflow canvas, improving user clarity and reducing confusion when interacting with node controls.

Health Assessment

• • Quick fix with minimal rework; single review and small scope.

Summary

Remove Jest support from testing packages

Eliminates dead Jest code paths, streamlining testing tooling and reducing maintenance overhead.

Health Assessment

• • The PR received a quick review with a single comment and no rework, but the merge was delayed by a week, likely due to scheduling rather than technical complexity.

Summary

Fix Google Docs default drive ID search

Prevents errors when switching Google Docs operations by adding a default drive ID, improving reliability for users.

Health Assessment

• • Fast turnaround with minimal changes and a single quick review, indicating low complexity and low risk.

Summary

Stop wrapping tool arguments in a value object

Fixes a runtime error caused by incorrectly wrapping object‑shaped arguments in the MCP client tool node, ensuring proper argument handling and dependency consistency.

Health Assessment

• • The PR received a quick initial review but required a change request after a few days, indicating moderate review friction; the fix was small to medium in scope and resolved a critical runtime error.

Summary

Correct documentation link

Fixes the in-app MCP documentation link to point to the correct MCP server documentation page, improving user experience.

Health Assessment

• • Quick fix with minimal changes, reviewed by AI, merged within 40 hours.

Summary

Fix ZIP64 decompression size limit error

Corrects false positive size limit for small ZIP64 archives, ensuring accurate decompression and preventing erroneous errors.

Health Assessment

• • The PR required several iterations and a dependency bump, but was reviewed quickly and merged after addressing the false positive size limit issue.

Summary

Fix Gmail node stringification of parameters

Ensures Gmail node handles non-string expressions by converting them to strings, preventing runtime errors and improving reliability.

Health Assessment

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

Summary

Fix expression isolate acquisition for node inputs

Ensures dynamic inputs in AI workflows resolve correctly in VM mode, preventing misreported connections and improving validation accuracy.

Health Assessment

• • Quick backport with minimal changes, no blockers, and fast review cycle indicate a low-risk, high-impact bug fix.

Summary

Fix expression isolate acquisition in node parameters

Ensures the VM expression engine correctly acquires an isolate during node parameter resolution, preventing failures in AI assistant debug actions and improving reliability for users.

Health Assessment

• • The PR was reviewed and merged within 17 hours, with minimal changes and no blockers, indicating a straightforward, low‑risk bug fix.

Summary

Add optional inference parameters to Bedrock node

Adds five optional inference parameters to the AWS Bedrock Chat Model node, enabling users to fine‑tune model behavior without switching providers. This enhances flexibility for users and improves model control.

Health Assessment

• • Fast review and single commit indicate straightforward change with minimal friction.

Summary

Release 1.123.64

This release updates package versions across the n8n monorepo, ensuring consistency and preparing for downstream deployments.

Health Assessment

• • Fast, automated release with minimal changes and no review needed.

Summary

Release 2.30.1

Bumps package versions and applies bug fixes and new features, delivering improved stability and new capabilities to users.

Health Assessment

• • Release merged immediately with no review delays, indicating a straightforward release process.

Summary

Release 2.29.8

Bug fixes improve data integrity and database resilience for production runs.

Health Assessment

• • Release merged immediately with no review, indicating an automated release workflow.

Summary

Backport bundle changes to release candidate

Backports recent bundle updates to the 2.29.x release candidate, ensuring compatibility and stability for the upcoming patch release.

Health Assessment

• • Backport completed swiftly with minimal review, indicating stable changes; large code churn but no blockers.

Summary

Add eval:langtracer-push for pushing eval cases

Enables automated pushing of Instance AI evaluation cases to a curated lang‑tracer suite, replacing manual per‑case calls and improving consistency and efficiency.

Health Assessment

• • Fast cycle time with minimal review iterations indicates efficient process; large code addition but well‑documented and tested.

Summary

Add RadioGroup component to design system

Provides a robust, accessible radio group component for the design system, improving UI consistency and developer experience.

Health Assessment

• • The PR was AI‑reviewed, leading to several iterations but a rapid first review. The large scope and extended cycle time indicate significant effort and potential risk for downstream integration.

Summary

Fix chat functionality on separate webhook server

Ensures chat works when n8n runs with a separate webhook server and the main process disabled, improving reliability for distributed deployments.

Health Assessment

• • Fast cycle time and minimal review indicate a straightforward bug fix with quick resolution.

Summary

Disable private credential actions in execution view

Prevents users from initiating credential connections or modifications while viewing a workflow execution in read‑only mode, enhancing security and consistency of the UI.

Health Assessment

• • Fast turnaround with a single commit and minimal changes; AI assistance streamlined the fix and regression tests were added to guard against regressions.

Summary

Fix RabbitMQ routing key resolution

Corrects routing key evaluation so each message uses its intended key, preventing all messages from using the first key.

Health Assessment

• • The PR was reviewed and approved within a few hours of opening, with only a single commit and minimal code changes, indicating a straightforward, low‑risk fix.

Summary

Add AI Agent node NDV controls

Provides users with advanced configuration options for AI Agent nodes, enhancing workflow flexibility and reducing setup friction.

Health Assessment

• • Long cycle time (137h) and multiple review rounds indicate significant rework; large code changes (over 5k lines) increase risk of defects.

Summary

Add sandbox memory limit variable for Merge node

Allows configuring sandbox memory limit for the Merge node, improving stability for large datasets.

Health Assessment

• • Quick review and merge within 2 days, minimal rework, stable change.

Summary

Fix chat endpoint on separate webhook server

Ensures chat functionality works when n8n runs with a separate webhook server, improving reliability for users interacting with chat widgets.

Health Assessment

• • Quick review cycle with AI bot approval, single commit, minimal rework, indicating a straightforward bug fix.

Summary

Prevent memory leak in MCP Server Trigger Node

Fixes a memory leak that caused OOM crashes during active MCP client polling by ensuring request frames are released and lowering idle session TTL, improving stability and resource usage.

Health Assessment

• • Quick review and merge within 19 hours, minimal rework, low risk.

Summary

Backport bundle to release candidate

Backport of changes from the main branch to release-candidate/2.30.x, ensuring new features and bug fixes are available in the next patch release.

Health Assessment

• • Merged within 0.9 hours with no review comments, indicating high confidence in the backport changes.

Summary

Add name-and-type and type-only credential matching modes for package import

Extends package import’s credentialMatchingMode to support fuzzy matching by name and type or type only, improving cross‑instance migration and credential resolution.

Health Assessment

• • The PR required multiple iterations and AI‑assisted review, indicating complexity; the large code changes and extended cycle time suggest moderate risk to delivery.

Summary

Stabilize debug mode E2E test

Fixes flaky E2E test for debug mode on failed executions, improving reliability of automated testing.

Health Assessment

• • Quick fix to stabilize a flaky test with minimal changes, low risk, fast turnaround.