I wanted WeChat as a lightweight interface for ChatGPT Work + Codex, so I built ClawBridge

I use ChatGPT Work and Codex on a Windows development PC, but once I leave the desk I usually don't want a full remote desktop or remote IDE. What I actually wanted was much smaller: check whether a ChatGPT Work task has finished; ask a follow-up question about the current local project; occasionally ask Codex to inspect the repo; do all of that from WeChat, which is already open on my phone. So I built ClawBridge for WeChat . The part I care most about is keeping the boundaries explicit: /work → observe local ChatGPT Work status/results /chat → continue the project conversation with ChatGPT /codex → explicitly run Codex CLI against a configured repo /chat never silently turns into code execution. When Codex is invoked, it runs as: text codex exec --sandbox read-only Source code, detailed logs, long outputs, browser state and runtime state all stay on the PC. WeChat is mainly the lightweight control and short-result surface. The current release is v0.1.1 . I also added: text python -m clawbridge doctor to check config, project roots, WeChat binding, Chrome / ChatGPT profile and Codex availability. Current status: Windows MIT licensed Alpha / early stage supervised real WeChat E2E tested 130 automated tests passing GitHub: github.com/samzhou1972/clawbridge-wechat I'm curious about one design question in particular: For a mobile companion to a local coding workflow, would you prefer a narrow interface like this, or something closer to a full remote IDE/session manager?