TinyMCE 8.7 is here, and it brings some of TinyMCE's newer capabilities together into one system, while opening a capability that self-hosted teams haven't had before. This release adds Custom Reviews, so you can define your own AI review prompts and control which ones your users see; connects TinyMCE AI to Suggested Edits and Revision History, so AI changes are tracked and labeled alongside human edits; and makes self-hosting generally available, which unlocks the biggest change: TinyMCE AI can now connect to MCP services like Jira. Here's what shipped in TinyMCE 8.7.
TinyMCE Custom Reviews: You decide which AI reviews run
The five built-in AI Review prompts (proofread, clarity, readability, length, and tone) are a useful first pass, but they don't know your brand voice, your compliance requirements, or anything specific to the app you're building.
Custom Reviews closes that gap. Integrators define their own review prompts and control what shows up in the end user's review list, so your chosen reviews are the ones available in your app. If your users need a domain-specific check, a brand voice guideline, a compliance pass, or a formatting rule that matters only in your product, you write it once and it's there for everyone.
Curious how to write one that holds up on a real document, not just a demo? The tinymceai_reviews reference covers the prompt schema and how to design for the way Review mode chunks large documents across parallel calls.
TinyMCE AI Collaboration tracking: The pieces finally talk to each other
Suggested Edits, Revision History, and TinyMCE AI have existed side by side for a while now. What they haven't done before is talk to each other. AI-generated edits ran through the same tracking as everything else, but reviewers had no way to tell whether a change was originated by AI or a person when they looked at a suggestion or a revision.
AI-driven suggested edits now carry a badge in the same review and version trail as human edits, so reviewers see one consistent revision history with the origin marked, instead of one undifferentiated trail.
You can also pause and resume Track Changes without losing suggestions already in progress. That's useful any time you need an AI pass to run without disrupting an active review.
See how the connection works in the Revision History documentation and Suggested Edits documentation.
Self-hosting is generally available, and it unlocks MCP
Self-hosting general availability is what makes the rest of this release possible. Before 8.7, TinyMCE AI only loaded from the Tiny Cloud CDN, so self-hosted deployments couldn't run it. Self-hosted TinyMCE AI can connect to MCP services such as Jira or Slack so in-editor AI can pull from the systems your app already runs on instead of working only from the currently open document.
MCP is available only for self-hosted deployments. Check out the on-premises deployment documentation for the setup details and screenshots.
Give TinyMCE 8.7 a try
If you've been waiting on MCP specifically, self-hosting GA is the release that makes it possible. If you haven't, Custom Reviews and AI Collaboration tracking are worth the upgrade on their own.
Read the full 8.7.0 release notes for the complete list of changes, then get your free 14-day trial of TinyMCE and see what 8.7 does in your app.
FAQs
What's new in TinyMCE 8.7?
Three things: Custom Reviews, so you can define your own AI Review prompts instead of using only the five built-in ones; AI Collaboration tracking, which connects TinyMCE AI to Suggested Edits and Revision History and lets you pause and resume Track Changes without losing suggestions in progress; and self-hosting general availability, which unlocks MCP for self-hosted deployments. The full release notes have the complete list.
Do I have to upgrade to TinyMCE 8.7?
No. 8.7 adds new capabilities on top of TinyMCE 8; it doesn't ask you to change how you're already using the editor. Check the 8.7 release notes before you upgrade, the same way you would for any release, and move when Custom Reviews, AI Collaboration tracking, or MCP solve a problem you actually have.
What is Custom Reviews in TinyMCE AI Review mode?
Custom Reviews lets integrators write their own AI Review prompts and control exactly which ones show up for end users, instead of relying only on the five defaults (proofread, clarity, readability, length, tone). See the tinymceai_reviews reference for how to write one.
Does TinyMCE AI now work with Suggested Edits and Revision History?
Yes, as of 8.7. AI-generated changes carry a badge in the same review and version trail as human edits, instead of being indistinguishable from them. You can also pause Track Changes for an AI pass and resume it without losing suggestions already in progress. Details are in the Revision History and Suggested Edits documentation.
Is MCP available on TinyMCE Cloud?
No, not yet. MCP connections only work for self-hosted and on-prem deployments in 8.7, and there are no cloud plans for it right now.
Do I need to self-host TinyMCE to connect AI to MCP services like Jira?
Yes. MCP is a self-hosted capability. If you're running TinyMCE Cloud today, you'd need to move to self-hosted to connect TinyMCE AI to an internal data source over MCP.
