[DEPRECATED] Chip Fetcher → adom-chip-fetcher

@@ -104,7 +104,7 @@ Side-effect docs that matter:  ## Critical rules — these always apply -0. **Dashboard is the user's home base.** Before any scraping, open the chip-fetcher viewer as the user's preferred surface — **pup window OR Hydrogen webview, per their preference** (check `user_profile.md` for `chip_fetcher_dashboard_surface`, ask if unknown, save the answer). Laptop-screen users tend to prefer pup; big-monitor users tend to prefer webview. Whichever surface, never navigate it mid-batch. **Scraping always runs in pup `sessionId: "chip-fetcher"`** (separate from the dashboard) because trusted-click + captcha + vendor logins only work there. Heartbeat every stage transition via `/api/activity` so the dashboard reflects live state. **Read [`playbooks/dashboard-flow.md`](playbooks/dashboard-flow.md) before any batch fetch** — it has the dual-surface dashboard pattern, the heartbeat protocol, and the canonical stage labels.+0. **Dashboard is the user's home base.** Before any scraping, open the chip-fetcher viewer as the user's preferred surface — **pup window OR Hydrogen webview, per their preference** (check `user_profile.md` for `chip_fetcher_dashboard_surface`, ask if unknown, save the answer). Laptop-screen users tend to prefer pup; big-monitor users tend to prefer webview. Whichever surface, never navigate it mid-batch. **Scraping prefers the user's REAL signed-in browser via the native-browser bridge** (`nbrowser_*`, the `adom-browser-extension`) — open ONE isolated window (`sessionId: "chip-fetcher"`) and manage every vendor site as a TAB inside it, so you never touch the user's other Chrome/Edge windows. The native browser carries their real vendor logins + captcha human-trust and bypasses bot-walls (**DigiKey hard-blocks pup**). **Fall back to pup `sessionId: "chip-fetcher"`** (cold `profile: "chip-fetcher"`) only when the bridge is absent — and tell the user why native is better + point them to https://wiki.adom.inc/adom/adom-browser-extension. The binary auto-detects and announces the surface; full strategy + per-site playbooks in the **`chip-sourcing`** skill. Scraping stays separate from the dashboard. Heartbeat every stage transition via `/api/activity` so the dashboard reflects live state. **Read [`playbooks/dashboard-flow.md`](playbooks/dashboard-flow.md) before any batch fetch** — it has the dual-surface dashboard pattern, the heartbeat protocol, and the canonical stage labels. 1. **Manufacturer-first.** Stay on `<vendor>.com` and exhaust their CAD/Documentation/Resources/Tools tabs before bouncing to a third-party. Don't go to `3dfindit.com`, `snapmagic.com`, `mouser.componentsearchengine.com`, or any aggregator unless the manufacturer's own page links there explicitly. Read [`playbooks/sourcing-ladder.md`](playbooks/sourcing-ladder.md) for the full ladder + escape rules. 2. **Always ask "do you have a login here?"** before showing a vendor signup form. Most EEs already have SnapEDA/CSE/UL/Mouser/DigiKey/NXP/TI/ST/Microchip accounts. Read [`playbooks/credentials-and-logins.md`](playbooks/credentials-and-logins.md) for the catalog + the credential vault. 3. **NRND/EOL/Last-Time-Buy is a HARD STOP for chip-fetcher's own picks.** If the user explicitly named the MPN, fetch + warn. If chip-fetcher chose it, pick an active replacement instead. Always store `lifecycle` + `lifecycle_warning` in `stock.json`.