Big fan of the Qualixar stack — SLM, slm-mesh, agent-amplifier, and the rest are all running across my home lab fleet (7 machines on a WireGuard mesh).
One limitation I keep hitting: both the SLM daemon and the slm-mesh broker hardcode 127.0.0.1 as the bind address. Since my machines are all on a trusted private WireGuard network (10.0.0.x), I'd love to run a single SLM instance on my always-on server and have every other machine connect to it — one shared brain, one shared mesh, no sync lag.
The fix would be simple env var support:
SLM_HOST=0.0.0.0
SLM_MESH_HOST=0.0.0.0
Right now I'm working around it by running slm and slm-mesh via npm on the server machine and pointing slm-mcp-hub at them, but native support would make deployment cleaner.
Happy to test it.
Big fan of the Qualixar stack — SLM, slm-mesh, agent-amplifier, and the rest are all running across my home lab fleet (7 machines on a WireGuard mesh).
One limitation I keep hitting: both the SLM daemon and the slm-mesh broker hardcode 127.0.0.1 as the bind address. Since my machines are all on a trusted private WireGuard network (10.0.0.x), I'd love to run a single SLM instance on my always-on server and have every other machine connect to it — one shared brain, one shared mesh, no sync lag.
The fix would be simple env var support:
SLM_HOST=0.0.0.0
SLM_MESH_HOST=0.0.0.0
Right now I'm working around it by running slm and slm-mesh via npm on the server machine and pointing slm-mcp-hub at them, but native support would make deployment cleaner.
Happy to test it.