← ClaudeAtlas

ha-options-flow-augmentlisted

Augments an existing Home Assistant Custom Integration with one generic config option retrofitted into its OptionsFlow — a post-setup setting stored in entry.options (never entry.data), with a typed selector, strings + translations, an entry-reload-on-change wiring, and a test. Non-destructive — existing options stay untouched. Activate on "add an option for X to the integration", "let the user configure the poll behaviour / a threshold / a toggle after setup", "retrofit an options flow", "füge eine Option für X hinzu", "erweitere den Options-Flow". Do not activate for greenfield scaffolding (ha-integration-scaffold, which ships the base options flow), config-flow setup patterns like tenant/reauth/reconfigure/zeroconf/oauth (ha-config-flow-augment), the coordinator poll-interval option specifically (ha-coordinator-add owns it), config-entry data migration (ha-config-entry-migrate), or deploying to a live HA instance.
nolte/claude-home-assistant · ★ 1 · AI & Automation · score 68
Install: claude install-skill nolte/claude-home-assistant
# HA Options Flow Augment Spec: `spec/claude/ha-options-flow-augment/en.md` (EN canonical) / `spec/claude/ha-options-flow-augment/de.md` (DE translation). This skill retrofits **one** generic, post-setup config **option** into an existing integration's `OptionsFlow` — the piece the initial scaffold's options flow cannot cover generically and that no other skill owns (the coordinator poll-interval is the one exception, owned by `ha-coordinator-add`). ## Why this is a skill, not an agent - **Human-visible augmentation surface** — the user describes an option and reads back the `OptionsFlow` change, the `strings.json`/translation entries, and the reload wiring; a skill keeps this on the visible command surface, like the sibling `ha-config-flow-augment`. - **Mid-flow interactivity** — the option key, its selector, its default, and whether a change must reload the entry are per-run dialogues the user confirms before generation. - **Bounded, inline generation** — one option plus its schema, strings, and test fits inline; no isolated agent context is needed. - Counter-dimension considered: the draft→validate loop could be an agent, but the selector/default decisions and the report belong in the user's working context; skill wins. ## When this skill activates Use this skill to add **one** post-setup option to an existing integration — a toggle, a threshold, a mode, a display preference — surfaced through the integration's `OptionsFlow` and read from `entry.options` at runtime.