Adding Markdown Autocompletion to Poznote via the CodeMirror 6 Editor Engine

[tlibouban]

Hi Tim,

First of all I'd like to thank you for your continious great work on providing us such a great easy to use, but in the mean time very complete, solution for our daily notes.

Context

Today, Markdown note-taking in Poznote relies on manual content entry in a simple source editor. This approach offers great flexibility, but it becomes time-consuming for users who regularly produce structured notes, documentation, or detailed reports.

Modern development environments have democratized autocompletion, which reduces keystrokes, speeds up writing, and limits syntax errors. This experience is now perceived as a productivity standard.

In Poznote, the lack of input assistance represents a missed optimization, even though Markdown is one of the primary editing modes. The technical observation is that robust autocompletion cannot be added as a mere plugin on top of the current input field: it is inseparable from a genuine editor engine. The request can therefore be reframed as follows: evolve the Markdown source editor into an engine capable of natively embedding autocompletion, automatic closing, and snippets.

Request

Implement a Markdown autocompletion feature within the Poznote editor, built on the Open Source editing engine CodeMirror 6.

The feature will cover:

• Automatic completion of common Markdown syntax (headings, lists, tables, links, images, code blocks, blockquotes).
• Contextual suggestions while typing (language inside code blocks, HTML tag completion, etc.).
• Quick insertion of predefined Markdown templates or snippets, with tabbable fields.
• Automatic closing and continuation of Markdown structures (pairs such as **, `, [], (); continuation of lists and blockquotes on line break).

Chosen Technical Solution: CodeMirror 6

The choice of CodeMirror 6 directly addresses Poznote's stack constraints (vanilla JavaScript frontend, Vite build already in place, privacy-first architecture with no CDN dependency, raw-source editing with preview):

• Framework-agnostic: integrates in vanilla JavaScript without imposing React, in line with the current architecture (React is used only to bundle Excalidraw).
• Modular and lightweight: tree-shakable packages, consistent with the project's "no heavy dependencies" philosophy — unlike a Monaco-type editor.
• Fully self-hostable: installed via npm and bundled locally with Vite, with no external calls, in keeping with the privacy-first model.
• MIT license: compatible with Poznote's MIT license.

Native coverage by the CodeMirror 6 modules:

Need	CodeMirror 6 module
Contextual suggestions + custom completion sources	@codemirror/autocomplete
Snippets / templates with tabbable fields	snippetCompletion (@codemirror/autocomplete)
Automatic pair closing	closeBrackets
Markdown syntax highlighting and parsing (GFM)	@codemirror/lang-markdown
Automatic continuation of lists and blockquotes	insertNewlineContinueMarkup
Language completion inside code blocks	@codemirror/language-data

In-house development then focuses on the completion sources specific to Poznote (table skeletons, link/image snippets, heading levels, and potentially cross-note references), with the engine handling the rest.

Scope and Points of Attention

• Nature of the work: replacing the current Markdown input field with a CodeMirror 6 engine. This is an editor refactor, to be estimated as such, and not a simple plugin addition.
• Syntax highlighting: CodeMirror highlights code within the editor via its parser; the existing highlight.js library can be kept for the preview. The scope of each must be clarified to avoid duplication.
• Consistency with the existing slash menu: keyboard autocompletion complements the "slash" menu already present, which remains relevant for touch/mobile use, where keyboard autocompletion is less useful.
• Possible fallback: if the refactor is not approved, a minimal solution remains possible (CodeMirror 5 + addons, or EasyMDE), at the cost of a maintenance-only software base.

Expected Benefits

• Time savings through fewer keystrokes when writing structured notes.
• Improved productivity for regular Markdown users.
• Fewer syntax errors and greater consistency across produced documents.
• Bringing Poznote closer to the standards of modern editing tools.
• Shared development and maintenance effort via a proven, widely adopted Open Source component (CodeMirror notably powers Joplin's Markdown editor).
• A differentiating feature and stronger retention of advanced users.

[timothepoznanski]

Ouch, this one is giving me a hard time, but I definitely understand the value of it. Thank you for the detailed suggestion and the kind words! 😊

[tlibouban]

Hi,

You‘re welcome, you deserve it quite well.

Regarding autocompletion, you said it was difficult but that it added value. Fair enough. But you didn't share your personal take on it.

What do you think?

Could a PR be on the table?

Do you want to do it on your end?

In any case, I like your goal of "putting users' data first."

I think this feature, delivered as an available option (but offered right from onboarding so it's tested by default), will find its users!

[timothepoznanski]

Hi,

First of all, thank you for asking what I think about it. I genuinely appreciate that.

To be honest, I spent most of today working on it, and I have to admit that it does make the Markdown editor feel nicer to use. Beyond autocompletion itself, CodeMirror also brings syntax coloring and a more polished editing experience overall.

That said, I'm still a bit torn.

One of my goals with Poznote has always been to keep external dependencies to a minimum. I want the project to remain as independent as possible and the codebase as simple as possible while still providing advanced features. Finding the right balance is important to me.

I also don't want to introduce excessive complexity or new sources of bugs that could make the project harder to maintain in the long run. My highest priority remains ensuring that users' data stays safe and that the project remains reliable.

Another reason is that I've always tried to keep the code approachable enough that, if needed, almost anyone could pick it up and understand it with the help of modern AI tools. Simplicity has value.

At the same time, I've received several requests over time asking for improvements to the Markdown editor, and I do understand that for people who write in Markdown all day long, features like autocompletion can significantly improve the experience. I want everyone to feel comfortable using the tool.

As for PRs, I'm always open to contributions. In this particular case, however, I started working on it immediately because your suggestion made a lot of sense to me, so I didn't really stop to think whether someone would submit a PR for it.

What I would suggest instead is releasing a beta version first so that you can try it in real-world usage and tell me honestly what you think. If the result feels right and proves stable enough, then we'll have much better information to decide how far we should take it.

Thanks again for the suggestion and for taking the time to challenge my thinking on this topic.

Tim

[timothepoznanski]

Hi,

I’ve published the beta version. Don’t hesitate to give it a try and let me know how it feels.

There are probably still a few bugs lurking around, even though I tested as many features as possible. This was quite a large change, so your feedback will be very valuable.

Thanks again for your support and for helping make Poznote better! 😊

https://github.com/timothepoznanski/poznote/releases/tag/6.23.0-beta

Important

This is a beta release so you must change the version in the docker-compose.yml