Container Runtime Issues
Container Runtime Not Running
Section titled “Container Runtime Not Running”Symptoms: “Cannot connect to Docker daemon” or container runtime not detected by scc doctor
Fix:
# macOS — start your runtimeopen -a OrbStack # If using OrbStackcolima start # If using Colimaopen -a Docker # If using Docker Desktop
# Linuxsudo systemctl start docker
# Verifydocker infoSCC works with Docker Engine, OrbStack, Colima, or Docker Desktop. It auto-detects the available runtime — Docker Desktop is not required.
Provider Image Missing
Section titled “Provider Image Missing”Symptoms: scc doctor shows “Image not found” for a provider, or scc start fails with an image error
SCC auto-builds provider images on first use. If the build failed or the image was removed:
# Check which images existdocker images | grep scc-agent
# Manually build the missing imagedocker build -t scc-agent-claude:latest images/scc-agent-claude/docker build -t scc-agent-codex:latest images/scc-agent-codex/scc doctor --provider claude (or --provider codex) shows the exact build command if the image is missing.
Container Won’t Start
Section titled “Container Won’t Start”Symptoms: Stuck at “Starting container…”
Fix:
scc stop --alldocker system prune -fscc start --freshSandbox Conflict
Section titled “Sandbox Conflict”Symptoms: “A sandbox already exists for this workspace”
| Situation | What happens |
|---|---|
| Stale (stopped) | Automatically replaced |
| Live (running) | SCC prompts: keep / replace / cancel |
| Force new | Use scc start --fresh |
Permission Denied
Section titled “Permission Denied”Symptoms: Docker permission errors on Linux
Fix:
sudo usermod -aG docker $USER# Log out and back in for changes to take effectContainer Cleanup
Section titled “Container Cleanup”Remove old containers:
scc prune # Remove stopped SCC containersscc stop --all # Stop all running SCC containersdocker container prune # Remove all stopped Docker containersProvider Auth Issues
Section titled “Provider Auth Issues”Symptoms: “Auth missing” or “Sign-in needed” in scc doctor output
Fix:
# Re-run setup to reconnect providersscc setup
# Or just start a session — SCC triggers browser sign-in automaticallyscc startCodex Auth Callback Failed
Section titled “Codex Auth Callback Failed”Codex uses localhost:1455 for its OAuth callback. If auth fails:
- Ensure port 1455 is free:
lsof -i :1455 - Kill any process using the port and retry
- Try closing and reopening the browser
Auth Readiness States
Section titled “Auth Readiness States”| State | Meaning | Action |
|---|---|---|
| Launch-ready | Auth + image present | Ready to go |
| Auth cache present | Auth OK, image missing | SCC auto-builds on start |
| Image available | Image OK, auth missing | SCC triggers sign-in on start |
| Sign-in needed | Nothing present | Run scc setup or start a session and let SCC bootstrap auth |
Network Policy Issues
Section titled “Network Policy Issues”web-egress-enforced not working
Section titled “web-egress-enforced not working”The web-egress-enforced policy creates an internal Docker network with a Squid proxy sidecar. If requests fail:
# Check if proxy container is runningdocker ps | grep scc-proxy
# Check the egress proxy imagedocker images | grep scc-egress-proxyIf the proxy image is missing, rebuild it:
docker build -t scc-egress-proxy images/scc-egress-proxy/