[baxyz]

Why

Sessions started from claude.ai/code on a devcontainer's Remote Control server run on Claude Code's default model. fajro sets a default in each container's Claude settings (devc, baxyz/fajro#22), but that only covers containers started by devc: an option of the feature carries it to every way of opening the container.

Change (claude-dev 1.4.1 → 1.4.2)

• New option remoteControlModel (string, default empty = Claude Code's default), saved in remote-control.env like the other options.
• claude remote-control has no --model flag, and the sessions it spawns inherit its environment: claude-rc exports the option as ANTHROPIC_MODEL right after loading its options.
• claude-rc status prints a model: line.
• README (options table, version history) and the AGENTS.md feature table.

Tests

• with_remote_control scenario sets remoteControlModel: sonnet; with_remote_control.sh checks the options file and claude-rc status (model: sonnet). Reading /proc/<loop>/environ would not work: it only shows the environment at exec, not a later export, and the server itself does not start in CI (no login).
• Locally: bash -n, shellcheck clean, and claude-rc status with a fake options file prints model: sonnet, or model: default without the option.
  Not verified: that a session spawned from claude.ai keeps ANTHROPIC_MODEL when the user picks another model in the UI (the UI choice is expected to win).

🤖 Generated with Claude Code

[github-actions]

✅ PR Validation Passed

All checks passed!

---

📋 Pipeline Status

	Job	Status
✅	🧾 Conventional Commits	passing
✅	🔖 Version Bump	passing
✅	🧪 Feature Tests	passing
✅	🧪 Scenario Tests	passing
✅	🐚 ShellCheck	passing

---

_{🤖 Generated by @helpers4 CI • 2026-11-08}