Pull Request Explorer

Summary

Quarantine submitted-workflow mocked credential setup test

Adds a Playwright test to quarantine.json to preserve a submitted workflow during mocked credential verification setup, reducing flaky test noise in CI.

Health Assessment

• • Quick turnaround with minimal changes and no blockers.

Summary

Configure episodic memory in n8n

Adds persistent episodic memory support, JSON configuration, and API endpoints for browsing memory entries, enhancing agent capabilities and user experience.

Health Assessment

• • Large-scale feature added with rapid review and minimal back‑and‑forth, indicating efficient collaboration but high complexity and potential impact on stability.

Summary

chore: Bump vm2 to 3.11.5

Updates the vm2 dependency to version 3.11.5, applying the latest security patches and bug fixes for the workflow engine.

Health Assessment

• • Fast cycle time and minimal review indicate a straightforward dependency update with low risk.

Summary

Disable Instance AI eval auto-trigger on PR review

Stops automatic AI evaluation runs on PR review, allowing manual control and reducing noise.

Health Assessment

• • Quick review and merge, minimal changes to CI workflow, low risk.

Summary

Add migration guidelines documentation

Adds comprehensive migration guidelines to help developers write and manage database migrations, improving consistency and reducing errors.

Health Assessment

• • The PR had a rapid review cycle but required several iterations to refine the documentation, indicating active collaboration but low technical risk.

Summary

Fix nodeGroups inclusion in source control export

Ensures node groups are preserved when exporting workflows to source control and CLI, improving consistency across environments.

Health Assessment

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

Summary

Surface pnpm install failures in setup-nodejs action

Preserves detailed pnpm error logs in CI, enabling faster triage and reducing false positives from timeouts.

Health Assessment

• • Fast review and merge with minimal changes, quickly resolving CI reliability issue.

Summary

Capture actual task durations in CI

Accurate task duration metrics are now captured in CI, fixing broken analytics. This unlocks reliable performance dashboards and supports cost-savings measurement.

Health Assessment

• • The PR was reviewed and approved within 17 hours with no additional commits, indicating a smooth process. AI-generated code was used to correct a critical CI metrics bug, improving dashboard accuracy.

Summary

Harden n8n sandbox service adapter

Improves robustness of the sandbox adapter by making destroy idempotent, adding reconnect recovery, merging environment variables, and enhancing error reporting, backed by comprehensive tests.

Health Assessment

• • Fast review and single iteration indicate a smooth process; robust tests provide confidence in the changes.

Summary

Clarify endpoints that accept binary uploads

Fixes misleading UI copy for Facebook Graph API node, clarifying which endpoints accept multipart binary uploads and adding tests to prevent regression.

Health Assessment

• • Quick review and approval after a single comment, indicating straightforward change.

Summary

Persist episodic memory entries

Adds persistence for episodic memory entries, sources, cursors, and cleanup, enabling advanced memory management for agents.

Health Assessment

Summary

Bump urllib3 constraint to >=2.7.0

Fixes high‑severity vulnerabilities by updating the urllib3 dependency constraint in the Python task runner, ensuring secure operation.

Health Assessment

• • Fast review and merge with minimal changes, addressing critical security issues promptly.

Summary

Update PostHog feature flag evaluation and filter events

Improves telemetry accuracy and reduces noise in analytics, ensuring reliable feature flag evaluation and compliance with PostHog updates.

Health Assessment

• • PR required significant rework over 10 days, but review was quick and AI review flagged no issues.

Summary

Clarify sticky([nodes]) does not add nodes to workflow

Documentation update to correct AI code-builder misunderstanding, ensuring developers know that sticky nodes must be added separately.

Health Assessment

• • Quick review and approval, minimal changes, low risk.

Summary

Fix editor signature

Corrects a bug in the editor by adjusting the function signature, ensuring proper node creation behavior.

Health Assessment

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

Summary

Add LLM eval server and credential rewrite

Enables a local LLM evaluation server with credential URL rewriting, controlled by a PostHog kill-switch, improving vendor SDK testing and debugging.

Health Assessment

• • PR had moderate iteration with three commits after the first review, but the overall cycle time of 50.4 hours remained within normal limits, indicating efficient resolution.

Summary

Fix core: prevent bogus workflow-id in submit-workflow tool

Ensures the instance‑ai builder agent correctly omits or uses valid workflow IDs, preventing failures and enabling self‑recovery within a single builder task, improving reliability of workflow creation.

Health Assessment

• • Quick fix with minimal changes, AI‑assisted review and summary, fast turnaround.

Summary

Fix builder workflow submit when missed

Ensures that when the workflow builder finishes without explicitly submitting, the orchestrator will detect and submit the workflow or record a clear failure, improving reliability and observability.

Health Assessment

• • The PR had a 53‑hour cycle with many merge commits but only a single quick review, indicating a focused but iterative process with high code churn.

Summary

Add integration tests for runtime-credentials module

Adds end-to-end integration coverage for the runtime-credentials backend module, ensuring secure credential handling and rule resolution.

Health Assessment

• • Fast cycle time with minimal rework; AI review helped catch issues early.

Summary

Refactor workflow resolver fallback helper

Adds a system-level credential resolver fallback, simplifying configuration and improving reliability for workflow executions.

Health Assessment

• • PR introduced a system fallback helper for credential resolution with a single review cycle, but the review took over a day, indicating moderate review friction.

Summary

Validate user-proxy tool outputs against api-types schemas

Ensures that user-proxy tool outputs conform to API type schemas, improving data integrity and preventing runtime errors.

Health Assessment

• • Quick review and merge indicate low complexity and minimal risk.

Summary

Prevent Insights chart axis from overlapping table heading

Fixes a UI bug where the Y‑axis tick label overlapped the table heading on the Insights page, improving readability when no data is present.

Health Assessment

• • Quick fix with minimal changes, resolved in under a day with no rework.

Summary

Add runtime skill loading foundation

Introduces a shared runtime skill loading framework that lets agents discover and load skills on demand, reducing context size and improving extensibility for future skill integrations.

Health Assessment

• • The PR was reviewed and merged within 22 hours, with AI review comments quickly identified issues. Despite a large code addition (~2,200 lines), the low review friction and single review round suggest a smooth integration process.

Summary

Migrate workflowsStore.workflowId usages in stores

This refactoring updates internal store references to prepare for removal of workflowsStore.workflowId, improving code maintainability and reducing coupling across the editor UI.

Health Assessment

• • Quick review and merge after minimal changes, indicating a smooth refactor with low risk.

Summary

Rename inbound-secrets module to runtime-credentials

Renames the backend module to align naming conventions, with no behavioral changes.

Health Assessment

• • The PR was reviewed and merged within 1.8 hours, with a single comment and no issues, indicating a smooth, low-risk refactor.

Summary

Improve Slack configuration UI for agents

Simplifies Slack integration for agents by removing OAuth support, adding guided app setup with configuration tokens, and enhancing the UI, improving security and user experience.

Health Assessment

• • The PR had a long review cycle (~44h) and required multiple commits after the first review, indicating significant rework. The large code churn (~2,100 lines) and many files suggest high complexity and risk.

Summary

Release 2.21.7

Version bump and critical bug fixes for core components, improving stability and performance.

Health Assessment

• • Fast merge with no review issues and minimal code changes.

Summary

Fix PostHog group identify call after init

Corrects telemetry integration to ensure group identification occurs after initialization, preventing missing analytics data for users.

Health Assessment

• • Quick review and merge, minimal changes, low risk.

Summary

Add multi-turn capability for IAI evals

Enables multi-turn conversation-based evaluation harness for Instance AI, improving test coverage and realism.

Health Assessment

• • PR required extensive review and multiple iterations, indicating complex changes and potential integration challenges.

Summary

Add sandbox name prefix and labels for Daytona

Enables easier identification and filtering of eval‑spawned sandboxes in Daytona, improving debugging and monitoring.

Health Assessment

• • Quick review and single commit indicate a smooth process.

Summary

Fix expression isolate for node executor

Fixes agent tool execution errors and reduces request load, improving stability and performance.

Health Assessment

• • Rapid review cycle with AI-assisted feedback enabled a quick, low-risk fix.

Summary

Add workflow validation tool for AI

Enables AI to detect workflow errors such as missing inputs, credential issues, and execution failures, improving automation reliability.

Health Assessment

• • Rapid approval with minimal review, indicating a straightforward implementation that was quickly validated by automated tooling.

Summary

Backport CI release tagging fix

Ensures experimental releases are correctly flagged as rc and only semver releases are marked as latest on GitHub, preventing release confusion.

Health Assessment

• • Merged immediately after creation with no review, indicating a straightforward change with minimal risk.

Summary

Update Bitbucket Node to use supported workspace endpoint

Ensures the Bitbucket Trigger node uses the correct workspace endpoint, preserving existing workflow compatibility. This fixes the broken workspace dropdown and maintains backward compatibility.

Health Assessment

• • PR addressed a specific API change with minimal code changes and a quick review, indicating an efficient process.

Summary

Fix scheduled-poll isolate acquisition failures via __emitError

Ensures that scheduled poll failures are surfaced as error executions, improving visibility and reliability for users.

Health Assessment

• • Backport with minimal changes, quick review, low risk.

Summary

Add episodic memory SDK defaults

Provides default episodic memory settings for agents, simplifying configuration and reducing manual prompt setup for users.

Health Assessment

• • The PR was reviewed quickly with AI assistance, but required several commits to address issues, indicating iterative refinement. The large code addition introduces significant new functionality while the fast review cycle helps mitigate risk.

Summary

Allow Unicode characters in execution metadata keys

Fixes metadata key validation to accept Unicode, enabling nodes with international names to work correctly.

Health Assessment

• • Quick review, minimal changes, low risk.

Summary

Release 2.21.6-exp.0

Bumps version numbers and updates changelog for the experimental release 2.21.6, enabling users to access the latest experimental features.

Health Assessment

• • Rapid merge with minimal changes indicates a straightforward release process.

Summary

Open workflow artifact immediately on edit

Ensures that when a user edits an existing workflow, the artifacts panel and canvas preview show the workflow instantly, reducing wait time and improving user experience.

Health Assessment

• • Quick review and merge with minimal iterations indicates high confidence and low risk.

Summary

Add project.id to OTEL workflow span

Adds a project.id attribute to OpenTelemetry spans for workflow execution, enhancing traceability and debugging for operations teams.

Health Assessment

• • Fast cycle time with a single AI‑reviewed comment and no major rework, indicating a smooth, low‑risk change.

Summary

Fix scheduled poll expression isolate acquisition

Ensures scheduled polling triggers acquire an expression isolate, preventing runtime errors under the VM expression engine and improving reliability of scheduled workflows.

Health Assessment

• • Fast review and minimal rework indicate a straightforward bug fix with clear test coverage.

Summary

Cache Node toolcache marker to improve CI performance

This change caches the Node toolcache marker file, preventing repeated downloads during CI runs. It improves build speed and reduces bandwidth usage.

Health Assessment

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

Summary

Remove grey background from AI timeline buttons

This change removes the distracting grey background from AI assistant timeline step buttons, improving visual clarity for users.

Health Assessment

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

Summary

Show clear error for tool node without Agent

When a tool node is executed without an AI Agent, the workflow previously crashed with a cryptic error. This PR replaces that with a clear, actionable message, improving user experience and reducing support tickets.

Health Assessment

• • The PR was reviewed quickly and merged without rework, indicating a straightforward bug fix with minimal risk.

Summary

Release 2.21.6

Patch release with bug fixes and version bumps across core and packages.

Health Assessment

• • Fast cycle time of under an hour, single review with no comments, minimal code changes, low risk.

Summary

Defer heavy idle-boot dependencies to reduce heap usage

Lazy‑loads large third‑party libraries so that n8n starts with ~3 MiB less idle heap, improving startup performance and lowering memory footprint for users who don't immediately use Instance AI.

Health Assessment

• • The PR was reviewed quickly and merged within 17.5 hours, indicating efficient collaboration. The large scope of lazy‑loading changes required careful review but was resolved in a single round.

Summary

Release 2.22.1

This release updates package versions and documentation, delivering bug fixes for webhook execution and AI vendor node validation, improving reliability.

Health Assessment

• • Release was merged quickly with minimal review, indicating a straightforward update.

Summary

Refactor typed-RPC dispatcher for node accessors

This refactor routes the remaining node accessor methods through the typed-RPC dispatcher, ensuring consistent handling and type safety. No user-facing changes are introduced.

Health Assessment

• • The PR was reviewed and approved within 20 hours, with minimal comments, indicating a smooth process.

Summary

Fix scheduled-poll isolate acquisition failures via __emitError

Ensures that failures to acquire execution isolates during scheduled polls are surfaced as error executions, improving visibility and reliability for users.

Health Assessment

• • Quick backport with minimal changes and a single review comment, indicating a straightforward fix.

Summary

Fix null pinData handling in instance AI

Ensures workflow executions don’t error when pinData is null by defaulting to an empty object, improving reliability.

Health Assessment

• • The PR was reviewed quickly and required only one round of changes, indicating a straightforward fix.
• • The small code change and lack of complex dependencies suggest minimal risk to the system.