Launching today
Wular Coder
Open-source always-on cloud agents, bring your own keys
7 followers
Open-source always-on cloud agents, bring your own keys
7 followers
Always-on agents you actually own. Open source alternative to Claude Tag, any model, bring your own keys and subs. Invoke from Slack, GitHub, Linear and more. PR reviews by your rules. Connect any MCP server via OAuth or API keys.
Free Options
Launch Team
Framer AI AgentsDesign and publish professional sites with AI
Promoted
Hey Product Hunt, I'm Muzamil.
I built Wular Coder because I wanted agents in my team's threads and github reviews without being locked into one platform and its models and still have the best experience.
In Coder Cloud you only pay for compute, counted only while a task is running: 300 CPU minutes a month free. Or bring your own runners (Vercel Sandbox, GitHub Actions, your own machine), at no extra charge.
It's beta, so I'd love to hear where it breaks.
The focus is on performance, usability and configurability with sane defaults. Most of the time, an agent you invoke anywhere should just work, with no need to visit the dashboard after the initial setup.
Agents can also change minor settings from the platform itself. Some ask for approval in the thread first, and the few critical settings stay dashboard-only, where the agent links you to the right page.
Right now agents run on a 1 vCPU / 2 GB machine only. Planning for them to be able to pick more powerful machines for running tests and other heavy workloads.
This was built on top of the initial version of Coder, now called Core. Core is a cross-engine subagent runtime, so you can spawn subagents in your harnesses using your existing subs from providers like Codex and Claude: https://github.com/muzam1l/coder#in-your-local-harness