# n8n releases 3.1.0 with fixes and new Agent features

published: 2026-11-03T15:05:26.385Z
topic: frameworks
source: https://github.com/n8n-io/n8n/releases/tag/n8n%403.1.0
url: https://aiagentnews.io/stories/n8n-releases-3-1-0-with-fixes-and-new-agent-features

n8n has published version 3.1.0, a pre-release with a long list of bug fixes and new features across its core, editor and nodes. The changelog highlights work on Agent versioning, the n8n Assistant, Apps, and memory recall.

The 3.1.0 release, dated 3 November, is marked as a pre-release on GitHub. The changelog lists many fixes, including changes to expression handling, queue-mode workers, outbound request guards, and the Agent token counting path.

New features include Agent version history and a publication timeline, separate publishing of saved Agent versions from drafts, and MCP tools for reading Agent session logs. The changelog also covers an on/off switch and environment variable for the n8n Assistant, a 25 MB limit for Agent knowledge file uploads, and improved episodic memory recall.

Several node changes are included, such as webhook signature verification for the Mailchimp and Venafi TLS Protect Cloud triggers, custom SSL certificate support for the Postgres node, and OAuth2 as the default for new Zendesk nodes. The chat hub module is turned off by default in this version.

## Why it matters if you build agents

If you build agents on n8n, Agent versions can now be published independently of drafts, and MCP tools can read Agent session logs.

The chat hub module is now off by default, so workflows that depend on it will need to turn it on again.

The Agent knowledge file upload limit rises to 25 MB, and the outbound request guard's default blocked IP ranges are expanded, which may affect workflows that call internal addresses.