adr-capturelisted

# ADR capture The expensive question in an enterprise codebase is not "what does this do" — the code answers that. It is "why is it like this", and the code cannot answer it at all. An ADR is where that answer survives the person who knew it. ## When a decision earns an ADR The test is **reach** — does it constrain work beyond this change? — and all three of these must hold: 1. **Hard to reverse.** Changing your mind later costs something real. 2. **Surprising without the context.** A future reader will ask why it was done this way. 3. **A genuine trade-off.** There were defensible alternatives and one was chosen for stated reasons. Miss any one and it is a ledger entry, not an ADR. Promoting small decisions is how a `docs/adr/` directory ends up with forty files and nobody reading any of them. Earns one: choosing a state management approach for an app; how apps share authentication; where a boundary between libraries falls; adopting or dropping a library; a deliberate deviation from an established pattern; a performance trade-off with a real cost. Does not: which variable name, which utility, anything a reviewer settles in one comment. A ledger entry gets promoted when it turns out others will have to live with it. ## The document `docs/adr/NNNN-kebab-title.md`, numbered sequentially and never renumbered — an ADR is referenced by number, so the number is an identifier. ```markdown # NNNN — <Decision in one line> Status: Accepted · Date: <ISO> · Deciders: <names> S