Albo

View saved version →

---
name: find
description: Search the user's Albo library — saved places, recipes, films, books, articles. Use when the user asks "what did I save…", "find that restaurant…", "what's on my wishlist", or similar.
---

# Find in Albo

Search the user's saved items with the Albo MCP tools.

## Steps

1. Translate the request into a `searchExtracts` query:
   - Free text → `query`
   - Category ("restaurants", "films") → `extractTypes` (use `getAvailableFilters` to see valid types)
   - "wishlist" → `isWishlisted: true`; "done/watched/visited" → `isDone: true`
   - Vibes/cuisines/themes → look up tags first with `searchTags`, then filter with `tagNames`
   - Ratings ("my favourites") → `minRating`
2. If the user names a collection, use `listCollections` to resolve it before applying its ID as a filter.
3. For location questions ("saved places near Shoreditch"), use `geocodeLocation` then `searchPlacesNearby`.
4. Use `searchImports` when the user asks for saved links, source imports, or markdown documents rather than extracted items.
5. Present results concisely: name, type, a one-line hook, and rating/tags when relevant. Offer to drill into any item (`getExtract`).
6. If nothing matches, loosen filters once before reporting no results, and say what you searched.

## Notes

- Paginate rather than dumping everything; ask before fetching more pages.
- For "what should I pick?" questions, prefer `getRecommendations`.

View saved version →

---
name: plan
description: Build an itinerary or plan from the user's saved Albo places — day plans, trip itineraries, date nights, food crawls. Use when the user asks to plan something using their saves.
---

# Plan with Albo

Turn the user's saved places into a concrete plan or itinerary.

## Steps

1. Clarify the basics if missing: where, when/how long, and any constraints (budget, dietary, with kids, etc.). One short question max — otherwise make sensible assumptions and state them.
2. Gather candidates:
   - `geocodeLocation` for the destination, then `searchPlacesNearby` for saved places in the area.
   - `searchExtracts` (query, `isWishlisted: true`, relevant `extractTypes`/`tagNames`) for wishlist items that fit.
   - `listCollections` if the user names a collection (e.g. "my Tokyo list").
3. Build the plan: group by day and neighbourhood to minimise back-and-forth travel, alternate meals/activities sensibly, and note opening-hours caveats as assumptions to verify.
4. Offer to save the finished plan back into Albo with `createDocument` so it shows up in the user's app.
5. If a collection for the trip exists, offer to add any newly relevant items with `addCollectionItems`.

## Notes

- Prefer the user's own saves over generic suggestions; only suggest non-saved fillers when there's an obvious gap, and label them clearly.
- Use `readCollectionMemory` for existing trip constraints. Use `saveCollectionMemory` only when the user states a durable collection-specific preference worth retaining (e.g. "vegetarian", "no museums").

View saved version →

---
name: save
description: Save a link, place, recipe, film, book or note to the user's Albo library. Use when the user shares a URL or asks to save/remember/bookmark something into Albo.
---

# Save to Albo

Save content into the user's Albo library using the Albo MCP tools.

## Steps

1. Work out what the user wants to save:
   - **A URL** (TikTok, Instagram, YouTube, article, recipe page, …) → use `saveUrl` so Albo's pipeline extracts the content automatically.
   - **A described item without a URL** (e.g. "remember this restaurant: Padella in London") → use `createDocument` to capture the details, or check `getAvailableFilters` for supported extract types first.
2. If the user mentions a collection ("save it to my Tokyo trip"), resolve it with `listCollections`, then use `addCollectionItems` after the save returns an item ID.
3. Confirm to the user what was saved and where. If extraction is asynchronous, say it may take a minute to appear in the app.

## Notes

- If a save deduplicates onto an existing import, say it was already saved instead of claiming a duplicate was created.
- If a save fails because the item already exists, offer to find it with `searchExtracts` or `searchImports`.
- Never invent extract types — check `getAvailableFilters` when unsure.
- Do not edit an existing markdown document unless the user clearly asked to replace its content.