The npm-installed Claude Code on spaceship-earth updated itself from the
pinned 2.1.304 to 2.1.308 right after its first login, so every play would
downgrade it and it would update again. New multica_claude_settings is
merged (recursively, not templated, since Claude writes the file too) into
~multica/.claude/settings.json.
provision-agent-host.yml sets DISABLE_AUTOUPDATER=1 and mirrors the model
mix already live on haunted-mansion: model sonnet, subagents haiku,
MAX_THINKING_TOKENS 4000, BASH_MAX_OUTPUT_LENGTH 20000, thinking off.
spaceship-earth: back on 2.1.304 and staying there; second run changed=0.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Found provisioning spaceship-earth:
- Login was not idempotent: `multica auth status` prints to stderr (0.6.5),
and the guard read stdout only, so every run re-logged in and restarted
the daemon. The guard now reads both streams; the status read is no_log
because it echoes a token prefix.
- `multica daemon start` / `daemon restart` fork the daemon inside the
caller's session (Ansible's SSH session scope), outside the systemd user
unit, which stays inactive: the same orphan seen on haunted-mansion on
2026-10-21. Start/restart now go through `systemctl --user`, after a
daemon-reload when autostart was (re)registered. A daemon found running
outside its unit is stopped first. The unit name is read from
`autostart status`, so --profile hosts resolve their own unit.
spaceship-earth: second run changed=0, daemon in multica-daemon.service.
Not yet run against haunted-mansion.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
rblundon