Pull Request Explorer

Summary

Fix thread suffix parsing for AI sessions

Resolves 500 errors for AI Builder sessions, restoring chat history and preventing data corruption.

Health Assessment

• • Quick review with minimal changes, but long cycle time due to merge timing; overall low risk and minimal rework.

Summary

Map user_id to owner_id for v2 Pipedrive

Remaps user_id to owner_id in v2 Deal and Activity requests, fixing API validation errors and allowing existing workflows to run without migration.

Health Assessment

• • The PR resolved a critical API mismatch with quick review turnaround and extensive tests, indicating a smooth and low-friction process.

Summary

Align credentials response with frontend type

Ensures the API returns complete credential data, preventing undefined values in client applications and improving reliability for workflow integrations.

Health Assessment

• • The PR was reviewed and merged within 8 hours, showing efficient collaboration, but required several commits after the first review to refine the API contract, indicating focused iterative improvement.

Summary

Fix lint rule to allow npm person emails

Corrects ESLint rule that incorrectly flagged valid npm author emails, improving lint accuracy for package metadata.

Health Assessment

• • Quick review and merge with minimal changes to lint rule, indicating low risk and high confidence in the fix.

Summary

Override workflow redaction policy at execution

Ensures instance-level redaction enforcement overrides workflow settings, preventing users from weakening data privacy controls.

Health Assessment

• • The PR introduced a new global redaction hook to enforce instance-level policies, required multiple iterations to resolve CI and type issues, but was reviewed quickly and merged after extensive testing.

Summary

UI for MCP servers in agents

Adds a user interface for managing MCP servers within agents, enabling selection and configuration of MCP connections directly in the editor.

Health Assessment

• • The PR required multiple iterations after the initial review, indicating complexity in integrating the new UI components. Review was performed by an AI tool (Cubic), which flagged issues across many files, contributing to a longer review cycle.

Summary

Use slugs instead of IDs for MCP registry servers

Switches server identification from mutable IDs to stable slugs, reducing race conditions and ensuring consistent server mapping across environments. This change improves reliability of MCP registry integration and simplifies test seeding.

Health Assessment

• • The PR required multiple iterations and a lengthy review cycle, indicating complexity in refactoring database identifiers and ensuring test coverage across environments.

Summary

Add sandbox service test container

Adds a sandbox service to the test container registry and simplifies the CI workflow, improving test reliability and reducing build time.

Health Assessment

• • The PR was reviewed quickly with minimal post-review changes, indicating a straightforward CI improvement. AI was used to review and generate a summary.

Summary

Fix nested sub-agent tool call handling

Ensures nested agent tool calls execute correctly, preventing workflow failures and improving reliability for users building complex automations.

Health Assessment

• • The PR required multiple iterations and a large code change to resolve nested agent tool call issues, indicating significant complexity. The review was quick but the overall cycle time was slow due to extensive testing and refactoring.

Summary

Add input validation for workflow and data table names

Adds Zod refinement to reject HTML markup in workflow and data table names, preventing persistence of malicious content.

Health Assessment

• • Fast review with AI assistance, minimal rework; long cycle time due to backport scheduling; overall low complexity and moderate risk for production release.

Summary

Fix recursive name collision in N8nToggle

Resolves a type‑checking error that blocked CI for node modules, enabling uninterrupted development and faster feedback.

Health Assessment

• • Quick fix with minimal changes, no rework, and no blockers.

Summary

Remove redundant memory string from OM

Eliminates unnecessary memory string from observation manager prompts, improving clarity and reducing payload size.

Health Assessment

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

Summary

Skip flaky sequential-interrupt checkpoint test

This PR temporarily skips a flaky test that blocks backend unit tests, allowing continuous integration to proceed while the underlying race condition is addressed. The skip prevents build failures and keeps the pipeline stable until the engine race is fixed.

Health Assessment

• • The PR was approved immediately with no rework, indicating low friction and minimal risk.

Summary

Move session state off workflows-store bridge

Refactor editor state management to use a dedicated workflow execution state store, improving modularity and reducing coupling.

Health Assessment

• • The PR required many post-review commits, indicating iterative refinement. Despite a quick initial review, the large scope and multiple changes raise moderate to high risk.

Summary

Refactor $getPairedItem routing via typed-RPC

Improves security and consistency by routing $getPairedItem through typed-RPC, adding schema validation and removing generic call channel. This internal refactor enhances trust boundaries without affecting user-facing behavior.

Health Assessment

• • The PR had a slow first review but minimal rework, indicating a straightforward internal refactor with no user impact.

Summary

Preserve underlying cause in webhook failure logs

Operators now see the full error chain in logs and Sentry, improving triage of webhook failures.

Health Assessment

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

Summary

Add tool to resolve node parameters

Enables AI assistant to inspect resolved node parameters during execution, improving debugging and automation accuracy.

Health Assessment

• • Fast review and merge, but many commits after the initial review indicate iterative refinement; overall low friction but moderate risk due to large code addition.

Summary

Refactor typed-RPC dispatcher for $evaluateExpression

This refactor routes $evaluateExpression through the typed-RPC dispatcher, ensuring consistent evaluation across VM and legacy engines. It preserves compatibility with community templates and AI workflow features, preventing potential breakage.

Health Assessment

• • The PR was reviewed quickly with no issues, indicating a smooth process. The refactor was straightforward and did not introduce new bugs.

Summary

Refine plan review UX with chat-input edits

Enhances the plan approval workflow by replacing inline textareas with a focused chat input, adding explicit actions, and improving visual feedback, leading to clearer user interactions and reduced confusion.

Health Assessment

• • Review took nearly 46 hours before first feedback, but no subsequent changes were required, indicating a smooth final approval. The PR introduces a large number of lines across many files, which could increase maintenance overhead.

Summary

Expand ESLint rule for toThrowError assertions

Improves linting accuracy by detecting error instance assertions in toThrowError, reducing false positives and ensuring code quality. This fix prevents linting errors across core and CLI packages.

Health Assessment

• • T
• • h
• • e
• •
• • P
• • R
• •
• • r
• • e
• • q
• • u
• • i
• • r
• • e
• • d
• •
• • m
• • u
• • l
• • t
• • i
• • p
• • l
• • e
• •
• • c
• • o
• • m
• • m
• • i
• • t
• • s
• •
• • a
• • f
• • t
• • e
• • r
• •
• • a
• •
• • q
• • u
• • i
• • c
• • k
• •
• • r
• • e
• • v
• • i
• • e
• • w
• • ,
• •
• • i
• • n
• • d
• • i
• • c
• • a
• • t
• • i
• • n
• • g
• •
• • i
• • t
• • e
• • r
• • a
• • t
• • i
• • v
• • e
• •
• • r
• • e
• • f
• • i
• • n
• • e
• • m
• • e
• • n
• • t
• • ,
• •
• • b
• • u
• • t
• •
• • t
• • h
• • e
• •
• • o
• • v
• • e
• • r
• • a
• • l
• • l
• •
• • c
• • y
• • c
• • l
• • e
• •
• • t
• • i
• • m
• • e
• •
• • r
• • e
• • m
• • a
• • i
• • n
• • e
• • d
• •
• • w
• • i
• • t
• • h
• • i
• • n
• •
• • a
• •
• • d
• • a
• • y
• • .

Summary

Add agent chat integration improvements

Enhances agent chat by adding Slack and Linear integration support, improving context handling, and streamlining workflow interactions, boosting productivity for users.

Health Assessment

• • The PR underwent rapid iteration with multiple merges and fixes within a day, indicating a complex integration effort. Despite a fast review cycle, the large code churn suggests high risk and potential for bugs.

Summary

Refactor NDV store to per‑workflow scope

Isolates NDV state per workflow, preventing cross‑workflow leaks and improving data integrity.

Health Assessment

• • Fast review (0.4h) and single commit indicate low complexity; the change is small to medium in scope and carries minimal risk to production.

Summary

Release 2.23.1

Version bump to 2.23.1, updating changelog and package versions.

Health Assessment

• • Quick release with minimal changes, no review needed.

Summary

Release 1.123.49

Bug fix that prevents connection leaks in webhook handling, released in a quick patch.

Health Assessment

• • Release merged within 18 minutes with no review friction and minimal code changes.

Summary

Refactor typed RPC dispatcher for node proxies

Refactors the typed RPC dispatcher to streamline node proxy interactions. This internal change improves maintainability and reduces error handling complexity, ensuring more reliable workflow execution.

Health Assessment

• • The PR underwent a single review after a 36‑hour delay, indicating moderate review friction. The refactor touches core runtime logic with a substantial code change, but the lack of multiple review rounds suggests confidence in the changes.

Summary

Fix Postgres Node SELECT empty array

Ensures SELECT queries that return no rows produce an empty array instead of a success flag, restoring correct workflow branching.

Health Assessment

• • The PR involved extensive regex refactoring and added comprehensive tests, resulting in a 12‑day cycle and moderate review friction.

Summary

Add MCP server support for agents

Enables agents to connect to MCP servers via a new schema and builder tools, expanding integration capabilities for users.

Health Assessment

• • PR added MCP server support with extensive tests; single review round indicates smooth integration and low friction.

Summary

Fix type-def synthesis for MCP registry nodes

Ensures Instance AI and Agent Builder can retrieve type definitions for synthetic MCP registry nodes, preventing node‑not‑found errors and improving runtime node handling.

Health Assessment

• • The PR resolved a runtime error for synthetic nodes with a single quick review and was merged after a week, indicating moderate complexity but straightforward resolution.

Summary

Release 2.22.5

Bumps version numbers across all packages, ensuring consistency and enabling the latest bug fixes and features for users.

Health Assessment

• • Release merged within minutes with no review needed, indicating automated release workflow.

Summary

Clear pin data on workspace initialization

Fixes regression where production executions incorrectly display pin data, ensuring correct execution data is shown.

Health Assessment

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

Summary

Refactor core to resolve $jmespath in-isolate for VM engine

Ensures $jmespath functions are resolved within the isolated VM engine, reducing host-callable surface and improving security and performance.

Health Assessment

• • The PR had a long cycle time but minimal review friction, indicating a straightforward refactor that was eventually merged after a few commits.

Summary

Allow users to disconnect their own dynamic credential connection

Users can revoke their own OAuth connections without affecting others, enhancing security and user control.

Health Assessment

• • The PR added a new user‑facing feature with significant backend and frontend changes, required multiple review cycles and AI‑assisted review, but was merged within a day with moderate review friction.

Summary

Render new empty state for quick-connect credentials

Fixes the credential setup UI for AI builder flows, ensuring quick-connect credentials display the correct sign-in button and improving user experience. This prevents users from seeing legacy dropdowns and streamlines credential onboarding.

Health Assessment

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

Summary

Show data redaction scope dropdown to unlicensed users

Adds a gated UI for data redaction settings, allowing unlicensed users to see options but requiring upgrade, improving UX consistency and preventing unauthorized changes.

Health Assessment

• • PR had a single AI-assisted review and a long cycle time due to multiple merge commits, but the change was straightforward and low risk.

Summary

Limit e2e-infrastructure triggers to dispatch and nightly

Reduces unnecessary CI runs, saving compute resources and speeding feedback loops.

Health Assessment

• • Fast merge with minimal changes, no review required, indicating efficient CI maintenance.

Summary

Disable turbo cache on test:changed tasks

Improves CI efficiency by preventing unnecessary cache writes for test:changed tasks

Health Assessment

• • The PR was merged quickly, indicating low review friction and a straightforward change

Summary

Add build-on-miss fallback to E2E docker image load

This change allows E2E tests to rebuild Docker images on cache miss, preventing hard failures. It improves test reliability and reduces false positives caused by cache eviction.

Health Assessment

• • Quick review and merge with minimal changes; AI-assisted code generation streamlined the update.

Summary

Track JWE enablement on credential events

Adds a jweEnabled flag to credential events and forwards it to PostHog telemetry, enabling measurement of JWE usage on OAuth2 credentials.

Health Assessment

• • Fast review and merge with AI assistance; single commit; minimal scope; low risk.

Summary

Show private credential connection state in credentials list

Adds UI to display connection status for private credentials, enabling direct OAuth flow and showing connected state.

Health Assessment

• • Merged within 34.5 hours after a single review, indicating a straightforward change with minimal risk.

Summary

Add project/folder actions menu

Users can now perform bulk actions on projects and folders directly from a new menu, improving workflow efficiency.

Health Assessment

• • The PR underwent extensive iteration with 10 commits after the initial review, indicating significant rework, but the review process itself was quick.

Summary

Correct OpenAPI schema for /source-control/pull

Aligns the public API spec with the actual implementation, eliminating mismatches that could cause integration errors for clients. This ensures developers using the API receive accurate request and response definitions, improving reliability and developer experience.

Health Assessment

• • Quick review and single commit indicate a smooth, low‑friction process.

Summary

Release 2.22.4-exp.0

This release updates package versions and documentation, including a new editor feature that surfaces the MCP access toggle on workflow cards, improving user workflow management.

Health Assessment

• • Fast, single-commit release with minimal changes, indicating a routine maintenance update.

Summary

Add fallback web search for agents

Adds native and fallback web search support for agents, enabling Brave/SearXNG utilities and improving builder reliability and memory defaults.

Health Assessment

• • The PR required extensive rework after the initial review, with multiple commits and comments over a 4‑day cycle, indicating significant complexity and potential integration risk.

Summary

Refactor $fromAI routing via typed-RPC dispatcher

Internal refactor to unify $fromAI aliases and streamline typed-RPC dispatching, improving maintainability without changing behavior.

Health Assessment

• • The PR was reviewed quickly with no issues, indicating a straightforward refactor with minimal risk. The merge occurred within a day, reflecting efficient collaboration.

Summary

Reduce Instance AI E2E flakiness

This PR reduces Instance AI E2E flakiness by improving proxy-backed replay tests and making the sidebar rename action more reliable.

Health Assessment

• • The PR has a low cycle time of 3.2 hours and a fast time to first review of 0.1 hours, indicating efficient review and merge processes.

Summary

Show default system resolver in workflow settings

Users see the system resolver as default when none selected, improving UX and preventing accidental credential pinning.

Health Assessment

• • Quick AI review with no issues, merged within 31 hours.

Summary

Add default OpenTelemetry trace filter for published workflows

Enables selective tracing of published workflows, reducing noise and improving observability.

Health Assessment

• • Fast review and merge with minimal rework; extensive test coverage ensures stability.

Summary

Direct-mount editor in AI artifact preview

Replaces iframe-based editor with directly-mounted editor for improved performance and usability.

Health Assessment

• • The PR has a large scope with 11 files changed, indicating significant changes to the codebase. The review process took 23.6 hours, which is relatively long, but the PR was ultimately merged after addressing the reviewer's comments.

Summary

Fix wrong description in runtime skills test

Corrects a broken test that incorrectly described runtime skills, ensuring test reliability.

Health Assessment

• • Quick fix with minimal changes, reviewed immediately by AI, no human review needed.

Summary

Stabilize runtime skill snapshot hashes

Ensures consistent snapshot names across CI and production, reducing drift and improving reliability for Instance AI deployments.

Health Assessment

• • Backport completed quickly with minimal changes, indicating low complexity and high confidence in stability.