Pull Request Explorer

Summary

Refactor: Fix small style nits in Agents

Improves visual consistency and text handling in the Agents editor, enhancing user experience.

Health Assessment

• • Quick review with no issues, small refactor, fast cycle time.

Summary

Refactor Instance AI workflow setup frontend

Improves maintainability and testability of Instance AI workflow setup, enabling faster feature delivery and reducing technical debt.

Health Assessment

• • The PR involved extensive refactoring across frontend and backend, requiring multiple iterations and a high number of review comments, indicating significant complexity and potential risk.

Summary

Fix RSS Feed Read Node to respect proxy settings

This change routes RSS feed requests through n8n's proxy‑aware HTTP stack, ensuring compatibility with corporate proxy settings. It also improves error messaging for connection failures.

Health Assessment

• • The PR was reviewed quickly with minimal changes, indicating a low‑complexity bug fix.

Summary

Track no results in code-builder search

Adds telemetry to detect missing node queries, improving AI workflow builder accuracy.

Health Assessment

• • The PR was reviewed quickly with no comments, indicating high confidence in the changes.
• • The moderate code churn and single review round suggest a low-risk, straightforward enhancement.

Summary

Improve MCP tools documentation usage

Enhances MCP tool documentation and encourages SDK reference usage, helping clients avoid errors and improve workflow creation efficiency.

Health Assessment

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

Summary

Add cla-signed label automation to CI

Automates CLA checks by adding cla-signed label when all contributors have signed, improving compliance and reducing manual review.

Health Assessment

• • Fast cycle time and single review indicate smooth integration; minimal code changes and automated workflow reduce manual effort.

Summary

Avoid Agent.close() deadlock in instance-ai web-research fetch

Fixes a deadlock that caused long-running fetches to hang, improving reliability and reducing build time.

Health Assessment

• • Quick review and single commit fix, minimal scope, low risk.

Summary

Allow same-domain redirects in instance-ai web research

Fixes redirect handling to prevent infinite retry loops, improving reliability of AI research tool and reducing build timeouts.

Health Assessment

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

Summary

Fix Resource Mapper types in SDK

Corrects loose types for Resource Mapper, ensuring workflow JSON validity and preventing errors when generating workflows via MCP.

Health Assessment

• • Merged within 2 hours with a single review cycle, indicating a straightforward bug fix with minimal friction.

Summary

Add deeplink pairing and connection handling

Enables native computer use via deep‑link authentication, improving user experience and security for desktop integrations.

Health Assessment

• • PR involved extensive code changes and multiple AI‑generated review comments, resulting in significant rework and a long cycle time.

Summary

Add location and attendees fields to calendar events

Adds location and attendees fields to Microsoft Outlook node’s calendar event create and update operations, enabling richer event details without breaking existing workflows.

Health Assessment

• • The PR introduced new fields to the Outlook integration, required multiple merges to resolve conflicts, and had a slow cycle time, indicating coordination challenges.

Summary

Show locked state for data redaction settings

Users without redaction permission now see disabled controls with a lock icon and tooltip, improving clarity and preventing unauthorized changes. This enhances user experience by making permission constraints visible.

Health Assessment

• • The PR underwent significant rework with multiple refactors and reverts, indicating complex changes and a need for careful review.

Summary

Add Instance AI thread provider

Enables per-thread AI runtimes, improving modularity and performance for AI workflows.

Health Assessment

• • The PR involved extensive refactoring across many files, leading to a slow review process and high scope, indicating significant effort and potential risk.

Summary

Add eval run detail loading and error states

Improve user experience with loading and error state handling for evaluation runs

Health Assessment

• • Multiple review rounds and commits indicate significant changes and potential for errors

Summary

Fix placeholder() to return string

Eliminates TypeScript build errors blocking AI agent workflow builder by simplifying placeholder model and credentials slot shapes, improving build stability.

Health Assessment

• • The PR required a lengthy review and several follow-up commits to resolve lint and placeholder handling issues, indicating moderate complexity and a need for careful testing.

Summary

Add OAuth2 authentication support to Figma trigger node

Enables OAuth2 credentials for the Figma trigger node, allowing secure webhook integration while preserving existing personal access token workflows.

Health Assessment

• • The PR was reviewed quickly with a single comment and approved after a few days, indicating a smooth review process and minimal friction. The addition of OAuth2 support is well-tested and does not introduce significant risk.

Summary

Add in-house CLA check workflow

Enables a self‑hosted CLA verification process that gates merges, reducing reliance on external services and improving compliance for contributors.

Health Assessment

Summary

Clarify agent builder prompt guidance

Improves user experience by clarifying prompt guidance and correcting memory capability statements.

Health Assessment

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

Summary

Align pairwise eval builder with production handover

Aligns evaluation builder with production handover, ensuring evaluation metrics accurately reflect real-world behavior, improving reliability and reducing false positives. This change enhances confidence in evaluation results and supports more reliable AI model deployment.

Health Assessment

• • The PR took 4 days to merge with a single review and one commit, indicating a moderate review cycle but a large code change that may impact evaluation reliability.

Summary

Persist agent chat draft and hide tool approval toggle

Improves user experience by preserving chat drafts across modes and removing an incomplete tool approval toggle, reducing confusion and preventing accidental usage.

Health Assessment

• • Quick review and minimal rework indicate a straightforward bugfix with clear intent.

Summary

Fix MongoDB node collection per-item resolution

Ensures MongoDB write operations correctly target per-item collections, preventing data misplacement and improving reliability for multi-collection workflows.

Health Assessment

• • The PR required extensive rework after a quick review, indicating complex changes to MongoDB node logic. Despite a fast initial review, the large number of subsequent commits suggests significant iteration before final merge.

Summary

Split SSO loader into feature-specific modules

Refactors the monolithic SSO loader into separate modules for SSO, SAML, OIDC, and provisioning, improving maintainability and testability.

Health Assessment

• • The PR involved a substantial refactor of the SSO loader, requiring multiple commits after the initial review, but the review was quick and no issues were found. The large code churn indicates significant changes but the process was efficient.

Summary

Fix YouTube Node Unlisted Privacy Status Typo

Corrects a typo that caused 100% failure when users set videos to unlisted, restoring functionality for affected workflows.

Health Assessment

• • Resolved typo in a single line with rapid review and merge, minimal risk.

Summary

Add n8n and workflow SDK versions to LangSmith metadata

Adds n8n and workflow SDK version tags to Instance AI traces, enabling filtering and grouping by release in LangSmith.

Health Assessment

• • Fast review and merge, minimal changes, no blockers.

Summary

Refactor workflow document store initialization

Replaces useWorkflowDocumentStore with injectWorkflowDocumentStore, reducing coupling and preparing for removal of workflow ID from workflows store, improving maintainability and future-proofing the editor.

Health Assessment

• • The PR involved extensive refactoring across 20 files with a large line change volume; AI-assisted review identified issues but no significant rework was required, resulting in a slow cycle time of 74 hours.

Summary

Export boolean CSV values as true/false

Ensures boolean columns in data tables are correctly exported to CSV, allowing accurate re-import and preserving data types.

Health Assessment

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

Summary

Add IP rate limiting to dynamic credential endpoints

Adds IP-based rate limiting to dynamic credential authentication endpoints, reducing abuse and protecting resources.

Health Assessment

• • Quick review and merge with minimal changes, indicating straightforward implementation.

Summary

Add liveness timeouts for Instance AI

Ensures stale work is cancelled consistently, improving reliability and user experience by preventing stuck threads.

Health Assessment

• • The PR required extensive refactoring and new timeout logic, resulting in a long review cycle and many commits before approval, indicating significant complexity and potential risk.

Summary

Fix node-sensitive output field redaction

Ensures workflow data integrity by preventing unintended redaction of sensitive fields, avoiding runtime errors in downstream nodes.

Health Assessment

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

Summary

Add observational memory storage foundation

Provides durable observation storage for agents, enabling thread‑scoped memory persistence and cleanup, which improves reliability and performance of workflow executions.

Health Assessment

• • The PR involved extensive changes across 20 files and 2,666 lines, leading to a slow cycle time of 138 hours. AI‑driven review helped surface issues early, but the large scope and multiple iterations indicate a high risk of integration challenges.

Summary

Scope credential resolution for agents

Aligns agent credential handling with workflow context, enhancing security and consistency for users.

Health Assessment

• • Rapid review and minimal back‑and‑forth indicate confidence in the changes, but the large code churn suggests significant refactoring of credential handling.

Summary

Add agents SDK telemetry hooks

Adds product‑agnostic telemetry support to the agents SDK, enabling consumers to capture AI SDK and LangSmith traces without embedding Instance AI‑specific behavior.

Health Assessment

• • The PR added over 1,000 lines of telemetry code and tests, received a quick initial review, and was approved after a few days, indicating a moderate review cycle but a large code change.

Summary

Skip unknown fixedCollection keys to avoid crashes

This PR prevents editor crashes when workflows contain unknown fixedCollection keys by silently skipping them, improving stability for API‑created workflows.

Health Assessment

• • The review cycle was unusually long (over 160 hours) but involved a single review round, indicating a straightforward fix that was delayed, likely due to low priority or scheduling constraints.

Summary

Remove unused memory backends

Eliminates dead SQLite and Postgres memory storage, simplifying agent memory implementation and reducing maintenance overhead.

Health Assessment

• • Fast review and merge with minimal rework, straightforward cleanup of dead code.

Summary

Generate service-specific OAuth2 credentials for dedicated MCP tools

Adds dedicated OAuth2 credentials for each MCP tool, enabling quick connect and preventing misuse in generic nodes.

Health Assessment

• • The PR underwent multiple iterations after a single AI-assisted review, indicating significant rework but a quick initial review cycle.

Summary

Fix Schedule Node hourly interval bug

Ensures scheduled workflows fire correctly for hourly intervals that do not evenly divide into 24 hours, preventing missed executions and improving reliability for users.

Health Assessment

• • The PR received a quick review but took over 5 days to merge, indicating a slow integration process likely due to frequent master merges.

Summary

Fix Wait Node expression resolution in custom HTML

Ensures expressions in custom HTML form fields are correctly evaluated, aligning Wait node behavior with Form node and preventing literal string rendering. This resolves a bug that caused unresolved expressions to appear as raw text, improving user experience.

Health Assessment

• • Quick review and approval with minimal changes; the PR fixed a specific bug with a small code change and test addition.

Summary

Add root redirect to instance AI when enabled

Users are redirected to the instance AI page from the root URL when the feature is enabled, improving onboarding and access to AI capabilities.

Health Assessment

• • Quick review and minimal rework indicate a straightforward change with low complexity.

Summary

Refactor Instance AI view into route-driven layout

Split monolithic InstanceAiView into layout shell and two route-children leaf views for better maintainability and scalability.

Health Assessment

• • The PR has a large scope with many files changed, indicating significant refactoring. The review process was iterative with multiple comments and revisions, suggesting some friction in the review process.

Summary

Update node popularity data

Automated weekly update of node popularity data used for sorting nodes in the node creator panel, ensuring the UI reflects current usage patterns.

Health Assessment

Summary

Fix MCP get_execution hang on circular references

Resolves indefinite hangs in MCP get_execution caused by circular references, ensuring reliable API responses for clients and improving system stability.

Health Assessment

• • Fast turnaround with a single review and no rework; AI review confirmed no issues; minimal code changes ensure low risk.

Summary

Fix missing sandbox workspace directories

Ensures sandbox workspace directories are always created, preventing 404 errors when listing workflows, improving agent reliability.

Health Assessment

• • Quick fix with minimal changes, resolved 404 issue for empty workflow lists, low risk to production.

Summary

Add webhook request verification to GitLab Trigger node

Adds verification of X‑Gitlab‑Token header to secure webhook requests, generating a secret token on creation and validating incoming requests, ensuring backward compatibility.

Health Assessment

• • The PR added a small number of lines and was reviewed quickly, but the merge was delayed by over 11 days, likely due to release scheduling rather than code complexity.

Summary

Fix waitTill propagation in scaling mode

Prevents sub-workflow failures by ensuring parent executions are correctly preserved during scaling, improving reliability of workflow resumption.

Health Assessment

• • Fast review and merge with minimal rework, indicating a straightforward bug fix.

Summary

Assign instance-ai CLI module to team

Update CODEOWNERS to assign the instance-ai CLI module to the instance-ai team, ensuring proper ownership and accountability.

Health Assessment

• • Quick review and merge with minimal changes; no blockers.

Summary

Disable chat during interactive agent choices

Disables chat input while interactive agent choices are pending, ensuring users respond to rendered cards instead of typing, improving UX and preventing confusion.

Health Assessment

• • Quick turnaround with minimal rework; review was almost immediate and no major issues found.

Summary

Add New agent to universal add menu

Adds a New agent entry to the global '+' menu and positions the Agents tab after Workflows, improving agent creation workflow for users.

Health Assessment

• • Quick review (0.1h) with a single comment, moderate scope (290 lines across 7 files), and AI-assisted summary generation suggest a smooth, low-risk feature addition.

Summary

Remove confirmation from browser connect/disconnect tools

Eliminates unnecessary confirmation prompts for browser connect and disconnect actions, improving user experience and reducing friction.

Health Assessment

Summary

Allow late browser connection after timeout

Enables users to reconnect browser extensions after initial timeout, improving reliability of browser-based workflows.

Health Assessment

• • Quick review and merge, minimal rework, demonstrates efficient collaboration.

Summary

Inline AI_NODE_SDK_VERSION to reduce memory

Reduces idle memory usage by inlining the SDK version, saving approximately 5 MB.

Health Assessment

• • Fast turnaround with no review issues; minimal code change.