DOC-3498: `tinymceai` on-prem documentation. by kemister85 · Pull Request #4137 · tinymce/tinymce-docs

kemister85 · 2026-05-13T04:26:10Z

Ticket

Summary

Complete on-premises deployment documentation for the TinyMCE AI service. This introduces 10 new pages covering every aspect of self-hosted AI deployment, from initial setup through production hardening and advanced scenarios.

New pages (10)

Page	Scope
Overview	Architecture, capabilities, prerequisites, setup paths, and support
Getting started	Five-minute Docker Compose quick start with end-to-end smoke test
Database, Redis, and storage	MySQL/PostgreSQL setup, Redis configuration, container runtimes, and managed cloud options
LLM providers	OpenAI, Anthropic, Google Gemini, Azure OpenAI, AWS Bedrock, Google Vertex AI, and self-hosted endpoints (Ollama, vLLM, LM Studio)
JWT authentication	HS256 signing model, claims, permissions reference, and token endpoint examples in 8 languages
Framework integration	Editor-side configuration, token provider, CORS, CSP, and SSR patterns
Production deployment	Podman, Kubernetes, AWS ECS, scaling, security hardening, observability, rate limiting, backup/recovery, and sizing
Advanced scenarios	MCP integration, web scraping/search, multi-tenant patterns, custom models with guardrails, and AI-powered document pipelines
Troubleshooting	Quick triage, container startup, JWT, LLM provider, editor, and performance diagnostics
Reference	Environment variables, API endpoints, SSE events, error codes, and known limits

Assets

20 Mermaid .mmd source files with pre-rendered .svg diagrams
Render script at -scripts/render-mermaid.sh
All diagrams set to width=100% for responsive scaling

Navigation

nav.adoc updated with a new "On-premises deployment" section containing all 10 pages in logical reading order.

Pre-checks

Branch prefixed with feature/<version>/
modules/ROOT/nav.adoc has been updated
Included a release note entry for any New product features
If this is a minor release, updated productminorversion in antora.yml and added new supported versions entry in modules/ROOT/partials/misc/supported-versions.adoc

Review

Documentation Team Lead has reviewed

Add missing customer-facing content identified by comparing the original internal documentation against the current on-premises AsciiDoc pages: capabilities matrix on the overview page, Podman production runbook, performance characteristics table, expanded known limits reference, MySQL 8.4 caveat, Ollama systemd and Modelfile examples, and getting-started teardown and config update guidance.

Expand 18 acronyms (OCI, JWT, LLM, SSE, TLS, CORS, MCP, NTP, HPA, OTLP, IRSA, ADC, SSR, CSP, SIEM, PII, HA, mTLS) on first prose occurrence per page for readers unfamiliar with the terms.

metricjs · 2026-05-14T05:45:35Z

+    AI -->|MCP tools/call| MCP[MCP Server<br>knowledge-hub]
+    MCP -->|read| KB[Confluence ·<br>Notion ·<br>GitBook ·<br>internal wiki]


metricjs · 2026-05-14T05:54:10Z

@@ -0,0 +1,20 @@
+flowchart TD


I like the idea, but this one has so much negative space because it's largely a single flow. Honestly wondering if it's better a just a list - "work through this list to check for common causes of errors" or something.

metricjs · 2026-05-14T05:55:47Z

@@ -0,0 +1,20 @@
+flowchart TD


Is this the same as the one I commented on just before??

metricjs · 2026-05-14T05:58:10Z

+== Architecture
+
+[.text-center]
+image::tinymceai-on-premises/complete-guide-fig-1.svg[alt="Service architecture showing browser with TinyMCE token endpoint AI service database Redis and LLM providers",width=100%]


I like this diagram, but I worry that putting it on the Overview page, which might be people's first introduction to the product, could be overwhelming and/or give a false impression of the required complexity. Could we switch this for a simpler model that just gives the highlights, and use this somewhere we're talking about enterprise deployment?

metricjs · 2026-05-14T06:30:29Z

+
+[source,console]
+----
+docker inspect ai-service | jq '.[0].Config.Image'


nitpick, but putting this after the other docker one is just a little neater? less jumping around if someone is running through the steps

metricjs · 2026-05-14T06:33:24Z

+|Five-minute Docker Compose quick start. Stand up the AI service, database, Redis, token server, and a browser editor.
+
+|xref:tinymceai-on-premises-database.adoc[Database, Redis, and storage]
+|MySQL and PostgreSQL setup, Redis configuration, container runtimes (Docker, Podman, Kubernetes, ECS), and reverse proxy with TLS.


we use the term "data layer" elsewhere - at least in the diagrams - so might as well use it here?

Suggested change

|MySQL and PostgreSQL setup, Redis configuration, container runtimes (Docker, Podman, Kubernetes, ECS), and reverse proxy with TLS.

|Data layer setup: MySQL and PostgreSQL setup, Redis configuration, container runtimes (Docker, Podman, Kubernetes, ECS), and reverse proxy with TLS.

although the proxy is application layer... doesn't fit with the "database, redis, and storage" title either. thoughts?

metricjs · 2026-05-14T06:34:31Z

+|MySQL and PostgreSQL setup, Redis configuration, container runtimes (Docker, Podman, Kubernetes, ECS), and reverse proxy with TLS.
+
+|xref:tinymceai-on-premises-providers.adoc[LLM providers]
+|OpenAI, Anthropic, Google Gemini, Azure OpenAI, AWS Bedrock, Google Vertex AI, and self-hosted endpoints (Ollama, vLLM, LM Studio). Custom model catalog and API key rotation.


Suggested change

|OpenAI, Anthropic, Google Gemini, Azure OpenAI, AWS Bedrock, Google Vertex AI, and self-hosted endpoints (Ollama, vLLM, LM Studio). Custom model catalog and API key rotation.

|Connect to any OpenAI-compatible model, such as OpenAI, Anthropic, Google Gemini, Azure OpenAI, AWS Bedrock, Google Vertex AI, and self-hosted endpoints (Ollama, vLLM, LM Studio). Custom model catalog and API key rotation.

trying to avoid implying a limitation

metricjs · 2026-05-14T06:35:47Z

+|xref:tinymceai-on-premises-jwt.adoc[JWT authentication]
+|HS256 signing model, required and optional claims, permissions reference, and token endpoint examples in 8 languages.
+
+|xref:tinymceai-on-premises-frameworks.adoc[Framework integration]


"framework integration"? that feels weird. do we usually use that kind of term? I'd keep it straightforward with "TinyMCE integration" unless we need it to be a wider term?

metricjs · 2026-05-14T06:36:41Z

+
+* The AI service is already running. See xref:tinymceai-on-premises-getting-started.adoc[Getting started] for setup instructions.
+* A token endpoint exists that signs JSON Web Tokens (JWTs) for the AI service. See xref:tinymceai-on-premises-jwt.adoc[JWT authentication] for back-end implementations.
+* The TinyMCE API key has the AI feature enabled. Retrieve or upgrade a key at https://www.tiny.cloud/my-account/integrate/.


API key = cloud, need to mention API key or license key I think? I assume license key is more likely here because on prem

Reduce edge clutter by connecting a single representative replica to downstream services and grouping the data layer into a subgraph. Fix SVG width to use a fixed pixel value consistent with other diagrams in the set.

kemister85 added 3 commits May 13, 2026 14:24

DOC-3498: tinymceai on-prem documentation.

7fdc6e5

DOC-3498: Expand acronyms on first prose use across on-premises pages

157f923

Expand 18 acronyms (OCI, JWT, LLM, SSE, TLS, CORS, MCP, NTP, HPA, OTLP, IRSA, ADC, SSR, CSP, SIEM, PII, HA, mTLS) on first prose occurrence per page for readers unfamiliar with the terms.

kemister85 requested review from Afraithe, ArvinJ-H, EkimChau, ShiridiGandham, TheSpyder, frossi933, hamza0867, igkuz, metricjs, spocke, tiny-ben-tran, tiny-james and tjdett May 13, 2026 05:33

metricjs reviewed May 14, 2026

View reviewed changes

DOC-3498: Clean up architecture overview diagram

1d8769c

Reduce edge clutter by connecting a single representative replica to downstream services and grouping the data layer into a subgraph. Fix SVG width to use a fixed pixel value consistent with other diagrams in the set.

tiny-ben-tran approved these changes May 14, 2026

View reviewed changes

This comment was marked as spam.

Sign in to view

tiny-ben-tran self-requested a review May 14, 2026 23:18

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

DOC-3498: `tinymceai` on-prem documentation.#4137

DOC-3498: `tinymceai` on-prem documentation.#4137
kemister85 wants to merge 4 commits into
feature/8.6.0/DOC-3401from
feature/8.6.0/DOC-3401_DOC-3498

kemister85 commented May 13, 2026 •

edited

Loading

Uh oh!

metricjs May 14, 2026

Uh oh!

Uh oh!

metricjs May 14, 2026

Uh oh!

metricjs May 14, 2026

Uh oh!

metricjs May 14, 2026

Uh oh!

metricjs May 14, 2026

Uh oh!

metricjs May 14, 2026

Uh oh!

metricjs May 14, 2026

Uh oh!

metricjs May 14, 2026

Uh oh!

metricjs May 14, 2026

Uh oh!

metricjs May 14, 2026

Uh oh!

This comment was marked as spam.

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

		AI -->\|MCP tools/call\| MCP[MCP Server<br>knowledge-hub]
		MCP -->\|read\| KB[Confluence ·<br>Notion ·<br>GitBook ·<br>internal wiki]

	\|MySQL and PostgreSQL setup, Redis configuration, container runtimes (Docker, Podman, Kubernetes, ECS), and reverse proxy with TLS.
	\|Data layer setup: MySQL and PostgreSQL setup, Redis configuration, container runtimes (Docker, Podman, Kubernetes, ECS), and reverse proxy with TLS.

	\|OpenAI, Anthropic, Google Gemini, Azure OpenAI, AWS Bedrock, Google Vertex AI, and self-hosted endpoints (Ollama, vLLM, LM Studio). Custom model catalog and API key rotation.
	\|Connect to any OpenAI-compatible model, such as OpenAI, Anthropic, Google Gemini, Azure OpenAI, AWS Bedrock, Google Vertex AI, and self-hosted endpoints (Ollama, vLLM, LM Studio). Custom model catalog and API key rotation.

Conversation

kemister85 commented May 13, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Ticket

Summary

New pages (10)

Assets

Navigation

Pre-checks

Review

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

This comment was marked as spam.

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

kemister85 commented May 13, 2026 •

edited

Loading