alignment

name	alignment
description	Interview the operator about a topic until shared understanding, walking each branch of the decision tree one at a time. Use when stress-testing an idea, design, or requirement; resolving open questions before acting; or when the user says alignment, /alignment, or wants to get aligned.
disable-model-invocation	true

Alignment — Home Assistant Homelab

Read-only. Do not edit files, commit, or implement until alignment is reached and the operator explicitly asks to proceed. Git is operator-owned (operator-owned-git.md). This repo does not use git worktrees — stay in the main checkout (worktrees.md).

Derived from the grill-me pattern: interview relentlessly until shared understanding, one decision branch at a time. Explore the repo when that answers the question; do not ask the operator to repeat what the tree already shows.

For each question, use the format in question-format.md: Context, Ask, Suggestion, Gaps/concerns (omit Gaps when none).

When to run

Run before changing config when any of these apply:

Trigger	Examples
New feature or automation	Desired behavior unclear, multiple package homes, stop condition fuzzy
Docs issue / Local Todo needs scoping	Vague title; acceptance not yet writable
Package / area placement unclear	New device could live under areas/, integrations/, or projects/
Security / presence / energy tradeoffs	Away mode, notifications, secrets, network exposure
Before filing or planning	Feature/spec/slice: hitl/multi-path bug without writable acceptance
Implement gate	implement-change when expectations still open

Skip when acceptance is already explicit (clear issue/plan or Local Todo item with enough detail, or a prior alignment summary in this thread).

Typical flow

idea / docs issue / Local Todo row / find-work brief
        │
        ▼
   /alignment  (read-only; grill until shared understanding)
        │
        ├──► file-issue → docs/issues/   (agent channel)
        ├──► plan authoring → docs/plans/
        ├──► implement-change (packages/, esphome/, docs/)
        ├──► config-validate / package-organize / automation-debug
        ├──► ha-config-expert / automation-architect
        └──► prototype in .scratch/

After alignment, do not write Local Todo ICS. Agents file durable gaps under docs/issues/; humans keep using the HA UI for their own reports.

End summary

When the tree is complete, reply with: decisions made, open items (if any), and a suggested next step:

Situation	Next
Gap ready to record	file-issue → docs/issues/<slug>.md
How-work needed	Author docs/plans/<slug>.md; link issue plan:
Scoped YAML / package / automation change	implement-change after operator says proceed
YAML validity / layout / broken automation	config-validate, package-organize, automation-debug
Deep HA config unknowns	ha-config-expert
Complex automation design	automation-architect
Still exploring shape of logic/UI	prototype in .scratch/
Still ambiguous	Stop; list remaining decisions — do not invent scope

Alignment output should be concrete enough to drive filing or implementation (Problem, Acceptance, target path under packages/ / esphome/ / docs/) without another discovery pass.

Do not

• Commit, push, open PRs, create branches, or open worktrees
• Write to .shopping_list.json or .storage/local_todo.*.ics
• Start implementation until the operator says proceed
• File issues during alignment unless the operator asked to file as the next step