allegro-pl-shoppinglisted

# Allegro.pl — site skill ## Workflow 0. **Consult the user's context** (context-first; mandatory, enforced by the runtime). After `asa run:start` (the context commands refuse to run without `run.json`) and before any search, for **each** need label, in this order: 1. `asa context:brief --need "<label>" --terms "<synonyms in the languages of the user's notes — RU, PL, EN — sizes in both × and x notation (180×200, 180x200), models, brands>"`. The runtime reads the knowledge stores of `CONTEXT_STORES` (an Obsidian vault inside its allow-list, the `shopping-profile/` files, plain folders), prints a digest with numbered snippets (`#1 date[!stale] file:line §heading [table columns] text (score)`) and writes the need into `.state/context-brief.json` (one brief per run, one entry per need; other needs keep their notes). Exit code 3 = no snippets: the digest says whether the store is mostly Cyrillic (add Russian terms) — add `--terms` and rerun, or go to step 3. 2. Read the digest and derive the specification (size, colour, quantity, brand, seller history, what was bought recently). Record every fact with the ids of the snippets it comes from: `asa context:note --need "<label>" --fact "…" --from "#3,#7"` (file:line is copied mechanically; a fact without `--from` is stored as an *unsourced* assumption). Mind the `!stale` marker and the `status` of a snippet: an old "choosing" note is not the same truth as a current "owned" line. 3. Record what the stores do not answer: `asa