[Feature] Have it as wasm lib for the browser

[suse-coder]

I would like to have a typescript app, that uses opencrabs as agent in the browser. Would it be possible to have it in the browser (sqlite or duckdb (wasm) work in the browser for opencrabs (wasm) also?

[adolfousier]

That sounds amazing bro, but let's clarify a few things.

You don't need a TypeScript app to run browser automation. OpenCrabs has built-in browser control via Chrome DevTools Protocol. Pure Rust, no Node.js or Playwright needed.

It can navigate, click, type, screenshot, and execute JavaScript in a real Chrome instance, both headless and headed (full visible browser). It can also run cron automations without human intervention.

OpenCrabs already handles LLM inference (via API providers like Anthropic, OpenAI, OpenRouter) and persistent storage (SQLite) natively. There's no need to compile the agent to WASM for that.

Where your TypeScript idea could make sense: OpenCrabs can execute external CLI tools. If you built a TypeScript tool with CLI control, similar to how Social Crabs works, OpenCrabs could call it as a tool and get results back. That's a much more practical approach than recompiling the entire agent to WASM.

One caveat: when using CLI providers (Claude, Qwen, OpenCode), the agent is limited to that CLI's own tools and memory. It can't call OpenCrabs' native browser tools. That's a genuine limitation I am actively looking at.

What's your actual use case?

If it's browser automation, we already have it. If it's building a custom tool that OpenCrabs orchestrates, that's very doable.

If it's running the full agent client-side in the browser, that's a much bigger project. Happy to explore what makes sense for what you're trying to build.

[suse-coder]

Use case would be to in a web site agentic use of my files (so it does not have to do it in the backend).

[suse-coder]

Not sure if that fits for the use case

[adolfousier]

What you actually want is a web interface where an AI agent can browse and organize your files. A WASM version of OpenCrabs will not solve this because browsers can not access your local files from a web app anyway.

Here is the right way to do it. Run OpenCrabs on your machine as a backend. Build a simple web page in front of it. The web page sends your commands and the agent handles all the file work on your machine.

This keeps your data and API keys safe on your machine. OpenCrabs already has every tool you need to read files, edit them, move them around, search through code, and run commands. The only missing piece is the web page to talk to it.

If you want to build that web frontend, the agent can help you with it. It can use its browser tools to test the UI while building it too.

What kind of files are you working with? What would opencrabs/agent do?