oms-ops

name	oms-ops
description	Maintain an oh-my-setting installation from chat: status, update, doctor, duplicate-skill cleanup, local snapshots, rollback, or uninstall. Use `oms-agent-harness` instead for repository task state.

oh-my-setting Operations

Run the local scripts for the user; do not send them to external connectors.

1. Orient with ~/.oh-my-setting/scripts/status.sh and use the canonical receipt owner it reports.
2. Use update.sh for updates, doctor.sh for install health, and skill-doctor.sh for duplicate/missing skill entries. The standard doctor automatically reports model/CLI capability drift when a provider is present; use --strict-diversity or --live-models only when the task needs those checks to become a blocking gate. --no-model-doctor is a recovery-only escape hatch, not a normal operating mode.
3. For legacy links, run cleanup.sh --dry-run, inspect its exact scope, then use --apply when cleanup is authorized. Never delete unrelated plugins, caches, regular files, or directories.
4. After update, repair, cleanup, or snapshot regeneration, run doctor.sh and report remaining warnings.

Machine and Slurm snapshot generators support --dry-run and --check; their output is private local state. A stale UI skill list may require a new agent session after the installed files are already correct.

Report the commands run, version/owner, and doctor result. Use compact prose; do not force an implementation-style report for a status question.