Release Apis

[Mullayam]

Also Please Release Api Based or Websocket based things so user can dpeloy on vps not only Install for host machine

[gajendraxdev]

Interested in this! the plan would be a self-hosted version you deploy on your own vps, keeping it on the roadmap.

[Mullayam]

@gajendraxdev yes im intereested but i dont know rust, what i build is https://terminus.enjoys.in in react_ts

[gajendraxdev]

@Mullayam I just checked out your project and it looks solid. The way you've handled multi-session SSH and the LSP proxying is exactly the kind of self-hosted architecture I'm looking for.

[Mullayam]

Hey @gajendraxdev , thanks!

But honestly, I’d suggest building on top of what I’ve already implemented + your own ideas, and push this to be even better than Terminus.

We can let users create a secure vault, sync hosts when they switch machines, and store data in the cloud. If you need a database or hosting infrastructure, you can use mine (Free of Cost).

SSH Features

1. AI suggestions / ghost text (similar to Terminus)
2. Auto error detection (AI detects errors/warnings and allows users to chat and fix them directly using open models)
3. Terminal sharing (with read/write permissions)
4. Recent commands / history
5. User-configurable settings (toggle options)

SFTP Features

1. Multiple SFTP sessions
2. Built-in file editor
3. Download files or entire directories as ZIP
4. Upload ZIP files

I'm using this since i discovered it + if you need and kind of help Please Feel Free to contact me

[gajendraxdev]

@Mullayam, thank you so much for the kind words and the generous offer to help, really appreciate it! 🙌

A few things to note on what's already in Zync:

• Built-in file editor powered by CodeMirror (syntax highlighting, works great for quick edits over SFTP)
• Download files or entire directories as ZIP

---

On LSP i thought about that but decided to skip it for now, the main reason is memory. running a full Language Server for each connected session can easily eat 200 - 1GB + of Memory. my plan is keeping it lightweight,
what your experience has been with this? Did you run into memory issues on Terminus, or did you find a way to keep it manageable?

On Cloud Sync me and @DanishSiraj thinking it differently then a central server, The plan is Zero trust serverless vault connections and ssh key encrypted locally and sync directly to their own Google Drive/One Drive via OAuth, And for team Who want more power they can host the backend to their own server and utilize sync on their one server, terminal sharing so on (Obviously it's not implemented yet 😅) but i will try to build it on top of what you alreaddy implemented

why this approach ?? Because A SSH Client manages sensitive production SSH keys and PEM files, I want to ensure users own their data, rather than requiring them to trust a third-party database,
and this is the exact reason why i moving away from traditional SaaS SSH clients (like Termius) and start building my own.

---

I'd love your help on the AI side! Your multi-provider setup (Groq, Mistral, OpenRouter, Gemini) looks solid. Would you be interested in contributing to improving Zync's AI command bar or the auto-error detection pipeline?

Also That genuinely means a lot, thank you! 😊The fact that you're already using it is the best kind of feedback 🙌🏻.

[Mullayam]

Also, if possible, please add support for multiple models/providers like if I want to use Groq or Mistral. I’m hoping that’s already supported or planned.

[gajendraxdev]

integrated already

it's working beautifully. I have two suggestions

• Icon Mapping: It would be a grate if the engine provided a way to get file-type icons (SVGs) based on the language ID.
• Thematic Data: Providing pre-built theme.json files for the standard Monaco color tokens (like Material, One Dark, etc.) would be huge.

---

Adding Groq and Mistral is a perfect fit for Zync’s BYOB model. I’m already setting up the provider-switching in the backend so users can toggle between Gemini, OpenAI, Claude, and local Ollama models and soon Groq/Mistral for that insane speed.

[Mullayam]

Clarification on Context Engine & Theming

To clarify, the context engine currently supports only linting and syntax coloring, along with 26 providers that integrate with Monaco Editor.

Theme Implementation

For themes, I’m using a structured approach like this:
Thematic data is defined in- a centralized module (https://github.com/Mullayam/terminus-web/blob/main/src/modules/monaco-editor/themes/index.ts)
Themes are loaded and applied dynamically in Monaco

I’m also working on a custom theme engine that converts Visual Studio Code themes into Monaco-compatible formats.

Additionally, I support themes from extensions by:

Using the VSIX registry
Extracting iconThemes and theme configurations from package.json
Converting and registering them within Monaco

This approach works well and has no issues so far.

Icon Strategy

For icons, I recommend handling them directly in the frontend (FE) instead of relying on Monaco/Context Engine.

Load SVG icons from external sources
Cache them using Cache Storage API to reduce latency
This avoids repeated network requests and improves performance

Icon Mapping

For file and UI icons, I’m using:
vscode-icons package
GitHub source: vscode-icons repository (raw SVG assets)
tabicon for tab-level icons

These are mapped based on file types and rendered in the UI layer (sidebar, tabs, etc.).

[gajendraxdev]

That make sense, Using the VSIX registry is way better than hardcoding everything, and caching the icons on the frontend is a much cleaner approach than bloating the engine with UI specific.

[Mullayam]

Themes will be covered in new version.
Also Some Deep like colors, document symbol, defination, tokenMonarch etc is going to be added.

[Mullayam]

Also, one more thing when you integrate new features, I’d like you to consider adding these:

1. Terminal AI Integration
   Allow users to open an AI chat from the sidebar or via a shortcut like Ctrl + I. The user can describe an issue, and the AI should generate the appropriate command and ask for confirmation before execution.
   The AI chat in the sidebar should have full context awareness, so users can interact naturally and execute commands directly from the chat interface.

2. Custom Header & UI Enhancements
   If you plan to add user login functionality, I’m not sure whether Tauri supports custom window headers like Electron (for close, minimize, maximize, tray, etc.). If possible, instead of placing elements in the main layout, add a custom header with profile, settings, and related controls similar to what Antigravity does.
   This way, the main layout can remain fully dedicated to SFTP and SSH functionality.

[Mullayam]

Today i Added This custom , Vscode , hosted a basic lsp on render

[gajendraxdev]

That VS Code-style setup looks very slick it definitely has that Pro feel!

In Zync, I’m currently opening the AI as a draggable modal (to keep it flexible across different views), but seeing your sidebar implementation makes me want to explore a 'Docked Mode' for terminal workflows. It looks much cleaner for long running sessions.

Also, Since we don't have a login/profile system yet, I'll probably focus the custom header on Settings and Connection Management to free up space

[Mullayam]

Just check this out, im looking this one in Zync,,

https://www.linkedin.com/posts/mullayam06_added-ai-agent-mode-to-my-terminus-web-based-activity-7443980859587698688-1fFR?utm_source=share&utm_medium=member_desktop&rcm=ACoAAEDf_-kB_X6rkG4GE_6ikheCM5xvFiH0R3w

https://editor.enjoys.in, Deployed on Vercel

[gajendraxdev]

Haha, consider it done! These features will be live in the next release, which also officially includes the new Monaco Editor integration. 🚀

[Mullayam]

Demo: https://monaco-lsp-hub.onrender.com (WIP)
Repo: https://github.com/enjoys-in/monaco-lsp-hub/tree/main
Few Language works very well, but due to internal LSP of monaco its not able to find out. Will Add and fix it soon

[Mullayam]

this is my custom editor that im using ,, LSP works Great Yeah just with some limitations

[gajendraxdev]

Hi @Mullayam
I just saw this message - checked the demo, it's really cool!

On Zync, I removed Monaco (I recently implemented it) and added a custom editor with your context-engine, but it's not fully stable yet.

I will explore monaco-lsp-hub and try to upgrade it if possible.

[gajendraxdev]

@Mullayam
I also added an AI side panel just like yours, it's live in v2.9.0. Has some issues, but I'll patch them in the next release.

recording-2026-04-05-170315_lSk6xIff.mp4