n8n 2.32.0

2.32.0 (2026-07-21)

Bug Fixes

• ai-builder: Honor eq @version conditions in generated node type definitions (#34232) (9a48bc5)
• ai-builder: Open agent chat preview on the same tab on instance AI (#34441) (6d56cee)
• ai-builder: Resurface channel edit modal on AIA agent builder's artifact mode (#34257) (0984a1d)
• ai-builder: Scope ToolIoView synth execution under a synthetic document store (#34588) (ab12ee5)
• ai-builder: Wire thread conversation history - with caps - for agents with slack channel (#34354) (8c1d17f)
• Allow devDependency imports in linter (#34335) (26caaa8)
• API: Fix broken public API playground server selection (#34077) (f24570f)
• AWS Bedrock Chat Model Node: Default guardrail version to working draft when not set (#33555) (013c34c)
• core, editor: Clarify credential setup outcomes and add requireUserSelection (#34268) (65e267b)
• core: Acquire expression isolate for MCP Trigger tool calls in queue mode (#34491) (eed8f3f)
• core: Add a per-attempt timeout to AI assistant service calls (#34341) (f191548)
• core: Add project ID and project name to log stream events (#33854) (a89817d)
• core: Alert on unusable Instance AI sandbox snapshots and gate the release on snapshot health (#34344) (080c3cc)
• core: Allow publishing end-user credentials with Send-and-Wait nodes (#34462) (c59f5d8)
• core: Bind dynamic credential tokens to a user ID (#34073) (fd261be)
• core: Cancel Instance AI tool runs promptly when generation is stopped (#34433) (4f8fef1)
• core: Cancel Instance AI workspace tools promptly on stop (#34551) (267fd85)
• core: Close the durable scheduler's record-then-dispatch effect boundary (#34014) (e22680b)
• core: Coerce AWS request header values to strings before signing (#34534) (d6f26cf)
• core: Count AI assistant workflow test runs as manual executions in statistics (#34145) (9842c2c)
• core: Create proxied Instance AI sandboxes from snapshots only (#34328) (29c5d72)
• core: Delete executions in batches when removing workflows, projects and folders (#34265) (0b809c9)
• core: Exclude in-flight Instance AI groups from history reads on non-driving mains (#34345) (50c3d88)
• core: Fix AI Agent not seeing tools connected through HITL tools (#34271) (bdff8f4)
• core: Fix change in width of all dropdowns in UI (#34249) (87cdf3d)
• core: Lint community packages against their provenance-attested source (#34214) (5f3d139)
• core: Persist MCP servers when credential setup is skipped (#34399) (7c7506d)
• core: Prefer direct tools for agent capabilities (#34466) (969a68e)
• core: Preserve agent builder question answers (#34414) (9882c5c)
• core: Preserve local credential fields absent from source contro… (#34173) (0a4ddde)
• core: Preserve node aliases when generating AI tool variants (#34162) (8284460)
• core: Prevent getIncomingData from throwing on a missing connection type (#34285) (9d05fbd)
• core: Prevent sub-workflow failures after transferring a deleted user's workflows (#33772) (7a56ade)
• core: Replace https agent for 'Ignore SSL issues' OAuth2 requests with one that respects proxies (#34234) (338f186)
• core: Require execution mode for MCP workflow executions (#34155) (79012f6)
• core: Resolve @n8n/expression-runtime to its CJS build under Node (#34463) (09f5ade)
• core: Retry transient AI assistant service errors during Instance AI environment setup (#34324) (5b7df71)
• core: Retry transient sandbox file writes (#34319) (80ea25b)
• core: Show a single error toast when a manual execution fails pre-flight validation (#34541) (8d1fc02)
• core: Sign SES requests with the ses service name (#34270) (2e31230)
• core: Skip reporting axios errors thrown as unhandled node errors (#34570) (33f1c36)
• core: Stop reporting quota-exhausted Instance AI runs to Sentry and isolate report request context (#34549) (6f0804e)
• core: Update error message on file write (#33833) (afe603f)
• Correct build-command quoting in standalone package release (#34258) (1301bba)
• Correct end-user credential validation, re-evaluation, and identity-hook recognition (#34094) (65da749)
• Default new agent MCP tool transport to httpStreamable (#34092) (9c3f7fa)
• Do not lose scenario results on a budget/timeout abort (#34419) (40c5b76)
• editor: Allow self-hosted OpenAI credentials for episodic memory (#34558) (e162724)
• editor: Auto-ungroup invalid groups on workflow save (#34456) (d5d3da6)
• editor: Avoid next prev buttons from being disabled (#33774) (a2fd7f7)
• editor: Decouple MCP redirect URI controls from env-lock flag (#34184) (64318a8)
• editor: Dissolve single-node canvas group when converting its nodes to a sub-workflow (#34334) (93e1b7f)
• editor: Improve Message an Agent panels (#34461) (e8b005e)
• editor: Include Chat, AI Assistant, and MCP settings under 'Instance settings: Manage' custom role permission (#34455) (fa5476c)
• editor: Keep dropdown tooltips above menus (#34547) (051eee1)
• editor: Link end-user credentials docs from credential modal (#34406) (09d1e96)
• editor: Make AI assistant quota-exhausted error states explicit (#33847) (4ef0d05)
• editor: Make selector dropdowns usable in Instance AI workflow setup (#33971) (487594c)
• editor: Polish pass on the custom instance roles (#34122) (297e252)
• editor: Prevent AI Agent from auto adding a chat connector on creation (#34259) (865035d)
• editor: Prevent users from clicking workflows within insights that they do not have access to (#34362) (e68fb3a)
• editor: Prevent workflow actions menu from being cut off (#34278) (8ba28dc)
• editor: Reconcile stale Instance AI run state on reload and stop (#34432) (8879b87)
• editor: Refresh editor checksum on workflow activation while editing (#34095) (7ebeb1e)
• editor: Refresh editor checksum on workflow deactivation while editing (#34315) (c83eab0)
• editor: Remove extra top spacing in sticky note markdown (#34116) (114c74d)
• editor: Rename agent tasks to schedules (#34326) (1ab30cb)
• editor: Restore hosted chat Markdown styles (#34176) (d6b9234)
• editor: Restore previous route when closing session trace (#34423) (e8922cb)
• editor: Restrict setup credential dropdown to pinned type (#34252) (454e0d3)
• editor: Show only the latest node as executing during manual executions (#34182) (6faca89)
• editor: Trigger options menu from node details header add button (#34212) (3b6af51)
• editor: Wrap long keyboard shortcut tooltip labels within max-width (#34135) (eb6a4a1)
• Even out workflow metadata divider spacing (#34132) (90f25e2)
• Extend orphan message cleanup to handle trailing orphans from memory trimming (#34385) (96264a0)
• Gate agent knowledge base table on AI Assistant proxy availability (#34366) (77880a3)
• Gmail Node: Handle multi-address Reply-To in the reply operation (#34312) (0b8ad18)
• Google Workspace Admin Node: Not able to reactivate the user (#33937) (f43d723)
• HTTP Request Node: Sign non-standard AWS endpoint hosts with the correct region and service (#34081) (0e8420b)
• Microsoft Excel Node: Resolve service principal target per input item (#34112) (afc8a87)
• Microsoft OneDrive Node: Resolve service principal target per input item (#34111) (578545c)
• Microsoft Outlook Node: Resolve service principal mailbox per input item (#34110) (831e054)
• Microsoft Teams Node: Validate Graph path IDs under all credential types (#34065) (e43a162)
• Microsoft To Do Node: Resolve service principal target per input item (#34109) (9b2479c)
• Postgres Node: Return DATE and date/time array columns as strings (v2.7) (#34050) (17eb819)
• Provide provider-accurate response shapes for evals (#34418) (1b173c7)
• QuickBooks Node: Initialize headers object in quickBooksApiRequest (#34291) (83dd194)
• Report an unverifiable unit as notVerified, not a silent pass (#34421) (7d6fd4c)
• Resolve modules on windows (#34167) (9f0a6fa)
• Salesforce Node: Pass ParentId when creating or updating a Case (#33775) (fffee46)
• Serialize date output in FTP, Set, IMAP and Unleashed nodes (#34205) (391deb9)
• Split Out Node: Properly handle if "Fields to split out" is not a string (#34148) (7998940)

Features

• Add showOnDeployment option, show warning for read file node on cloud (#34417) (f470230)
• Add typed seed data tables with rows on execution scenarios (#34420) (02af123)
• Adding agents tracing environment variable (#34401) (b42aeee)
• ai-builder: Allow AIA agent builder to publish and unpublish first-class agents (#34368) (84ee063)
• ai-builder: Help AIA remove / adjust agent channels (sp Slack) via chat (#34241) (f851863)
• ai-builder: Refresh Instance AI agent artifact UI on channel setup success (#34465) (2955d75)
• ai-builder: Send agent preview chat context to AIA chat (#34308) (9fad2ba)
• API: Add endpoint to retrieve SAML SSO configuration (#34168) (e76b345)
• API: Add endpoint to set the SAML SSO configuration (#34224) (cf8337a)
• API: Add public API endpoints for log streaming destinations (#34251) (4a651dd)
• API: Add public API endpoints to trigger and cancel evaluation test runs (#33979) (18b34ca)
• API: Allow creating a folder in the calling user's personal project (#34091) (571e642)
• API: Support placing workflows in folders via parentFolderId (#32696) (a844d03)
• AWS Bedrock Chat Model Node: Add Timeout option (#34072) (a4738ea)
• core: Add per-workflow execution data write bytes metric (#34425) (f7496a5)
• core: Add tracing, credit metering, and telemetry for the builder sub-agent (#34151) (be35343)
• core: Add variable bundling to package export (#34040) (b637582)
• core: Add workflow requirements to package exports (#34217) (581e490)
• core: Adopt "n8n credits" wording and default to it in Instance AI (#34426) (263ec69)
• core: Enable n8n Connect usage in n8n MCP (#33723) (4e4283b)
• core: Enable prompt caching for the agent-builder sub-agent (#34352) (a9357ef)
• core: Include error workflows as export dependencies (#34350) (7d23168)
• core: Prefer MCP servers over node tools when available (#34458) (078ee55)
• core: Support agent preview handoff in Instance AI (#34171) (5a3394a)
• core: Unify output schema mock generation and verify AI Assistant workflows with any trigger type (#33896) (d33c282)
• editor: Add descriptions to canvas groups (#34219) (8477e58)
• editor: Add Mine/All tabs, search and filters to MCP connected clients (#33857) (a2c19e4)
• editor: Add new Tags Input component to @n8n/design-system (#34075) (bac57c4)
• editor: Add per-case compare tabs to eval collection compare view (#33880) (dde77cb)
• editor: Add scope selection to MCP OAuth consent screen (#33710) (c3ca9a3)
• editor: Add sticky support for canvas groups (#34317) (0fb07ca)
• editor: Collapse AI Assistant thinking traces into thinking blocks (#34274) (9744625)
• editor: Improve sub workflow extraction (#34169) (318925a)
• editor: Rename N8nActionBox to N8nEmptyState and add icon-cards variant (#34407) (a6b5462)
• editor: Save canvas group description pinned state in local storage (#34220) (ecff317)
• editor: Show access and client details in the connected MCP clients list (#33714) (0ca24ba)
• editor: Show per-scope tool details on the MCP consent screen (#33783) (cc515a7)
• Gmail Node: Add advanced HITL configuration for Send and Wait (#34255) (6194110)
• Grist Node: Use a single Grist URL field in the credential (#34190) (d6d7d93)
• Mailgun Node: Add reply-to, custom headers, and tags (#28558) (b9199e5)
• Slack Node: Add advanced HITL configuration for one-tap approval (#34123) (d30e7b0)
• Telegram Node: Add advanced HITL configuration for one-tap in-chat approval (#34009) (6db46bc)
• Validate agent configuration before publishing (#34452) (5cb674e)

---