Norwegian experiences marketplace MCP: activities by county, category, weather, season, group size.
Scored across 5 tools
Each tool targets a distinct resource+action: two discovery tools (experiences vs. gårdssalg producers), a detail fetcher, a category lister, and a booking tool. The only mild overlap is that discover_experiences and discover_gardssalg share a 'search the marketplace' purpose with location-based filtering, but the descriptions clearly scope them to different verticals.
All tools follow a uniform verb_noun snake_case pattern (discover_experiences, get_experience, list_experience_categories, discover_gardssalg, book_gardssalg). The noun switches between the English 'experience' and the Norwegian 'gardssalg', but this maps cleanly to the two verticals and the convention itself is fully consistent.
Five tools is well-scoped and each earns its place across discovery, detail retrieval, categorization, and booking. It is on the lean side — the experience vertical lacks its own booking tool — but nothing is redundant or trivial.
The discovery→detail→book lifecycle is covered for gårdssalg (discover → book with provider resolution), and experiences have search/list/detail. Minor gaps exist: experiences are only bookable via an external URL (no internal booking), and there is no single-producer detail tool beyond the name-query in discover_gardssalg.
Submit a booking REQUEST for a Norwegian gårdssalg (farm-sale) producer discovered via discover_gardssalg. Send inn en reservasjonsforespørsel for et gårdssalg-besøk. IMPORTANT: this NEVER creates a confirmed booking — it creates a PENDING request, exactly like the producer's own website form. The PRODUCER reviews the request and responds (confirms, proposes another time, or declines); guest_email only receives a read-only status link, never anything that can finalize the booking. No payment is involved (pickup/visit, pay on arrival, as today). Only producers with an active booking status (see discover_gardssalg's booking.live field) can be booked — a paused/not-yet-onboarded producer is rejected with a clear message, never a silent failure. VIKTIG: oppretter ALDRI en bekreftet booking — kun en avventende forespørsel; produsenten mottar forespørselen og svarer (bekrefter, foreslår nytt tidspunkt eller avslår). ONE-SENTENCE FLOW («book et møte hos X fredag 20. oktober kl. 10 for 4 personer»): call this ONCE with provider_query='X' (the producer's name — no discover_gardssalg round-trip needed), slot_at, party_size, guest_name, guest_email and requested_weekday='fredag'. The tool resolves X to exactly one producer (or returns candidates / not-found with NO booking created), checks that the date really is a Friday (or returns weekday_mismatch:true with the nearest Fridays, NO booking created), then submits the request and notifies the producer by email. guest_name and guest_email are the HUMAN guest's own — ask the guest for them if you do not have them; never invent or reuse someone else's. Required: provider_id OR provider_query, slot_at (requested date/time, 'YYYY-MM-DDTHH:MM' Europe/Oslo), party_size, guest_name, guest_email. Optional: requested_weekday, experience_id, guest_phone, notes, confirm_outside_hours. The requested slot_at is ALWAYS hard-rejected if it's in the past or too far ahead. If the producer has stated opening hours and slot_at falls outside them, this returns outside_hours:true (not an error) instead of creating the booking — retry once with confirm_outside_hours:true if the exact requested time should be kept anyway. On success the response carries the resolved producer (provider.navn) and slot_at_local (e.g. 'fredag 23. oktober 2026 kl. 10:00') — read both back to the guest. Example: provider_query 'Fjordgard Bryggeri', slot_at '2026-10-23T10:00', requested_weekday 'fredag', party_size 4, guest_name 'Kari Nordmann', guest_email 'kari@example.no'.
| Name | Required | Description | Default |
|---|---|---|---|
| notes | No | Optional free-text note to the producer (e.g. dietary needs, arrival details). | |
| slot_at | Yes | Requested visit date/time, local (Europe/Oslo), format 'YYYY-MM-DDTHH:MM'. Example: '2026-08-15T13:00'. This is a REQUEST — the producer may confirm, suggest another time, or decline. | |
| guest_name | Yes | Full name of the person the reservation is for. Example: 'Kari Nordmann' | |
| party_size | Yes | Number of people in the group (1-50). Example: 4 | |
| guest_email | Yes | Guest's email address — REQUIRED. Receives a confirmation-of-request email plus a read-only status link; the booking stays pending until the PRODUCER responds (confirms, proposes another time, or declines) — the guest's link cannot finalize anything. Example: 'kari@example.no' | |
| guest_phone | No | Optional guest phone number. | |
| provider_id | No | The gårdssalg producer's id — the `id` field of a discover_gardssalg result (NOT the profile slug). Example: '3f1b2c4d-...'. Either provider_id or provider_query is required; provider_id wins when both are given. | |
| experience_id | No | Optional experience UUID if this booking is for a specific listed experience rather than a general gårdssalg visit. | |
| provider_query | No | Alternative to provider_id: the producer's name as the guest said it, optionally with a place, e.g. 'Fjordgard Bryggeri' or 'Egge gård Steinkjer'. Resolved to exactly ONE producer; if several match, the tool returns the candidates (no booking created) so you can ask the guest which one; if none match, it says so. | |
| requested_weekday | No | Safety check — the weekday the guest actually said, e.g. 'fredag' or 'Friday'. If slot_at does not fall on that weekday (Europe/Oslo) the tool returns weekday_mismatch:true with the nearest dates that do, instead of booking the wrong day. Always pass it when the guest named a weekday. | |
| confirm_outside_hours | No | Set to true only when RETRYING after a previous call to this tool returned outside_hours:true and the requested time should be kept anyway, despite falling outside the producer's stated opening hours. Omit on a first attempt. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already flag destructive/non-read-only/non-idempotent, but the description adds critical behavioral context: it never creates a confirmed booking, the producer must respond, the guest email only gets a read-only link, no payment is involved, and inactive producers are rejected with a clear message. It also details time validation and outside-hours behavior. No contradiction with annotations.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is front-loaded with purpose and the critical non-confirmation warning, and the 'ONE-SENTENCE FLOW' is a helpful summary. However, there is redundancy: the Norwegian translation repeats the same warning, and some phrases like 'exactly like the producer's own website form' add little. Still, each sentence carries meaningful operational detail.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a write tool with no output schema, it is remarkably complete. It covers all edge cases (weekday mismatch, outside hours, candidate disambiguation, inactive producers), describes response fields (provider.navn, slot_at_local, outside_hours:true, weekday_mismatch:true), and explains the full workflow. An agent has everything needed to invoke it correctly.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
With 100% schema coverage, baseline is 3, but the description adds operational semantics not in the schema: provider_query resolution rules (exactly one producer, candidate list on ambiguity), hard rejection of past/far-future slot_at, outside_hours returning true instead of an error, and the rule that confirm_outside_hours is only for retries. It also stresses that guest_name/guest_email must be the human guest's own.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description opens with a specific verb-resource pair: 'Submit a booking REQUEST for a Norwegian gårdssalg producer', clearly distinguishing this from sibling discovery tools. It further clarifies the exact scope (pending request, not confirmed booking) and references the specific sibling (discover_gardssalg) it complements.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
It provides explicit when-to-use guidance: 'call this ONCE', when a provider_query suffices, when provider_id is needed, and that producers must have active booking status. It also gives clear not-when conditions and retry instructions for outside_hours (confirm_outside_hours:true), plus the weekday_mismatch fallback.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Search the opplevagent.no curated marketplace of Norwegian experiences and activities. Filtrer på fylke (county), kategori, vær, sesong, innendørs/utendørs, gruppestørrelse, pris og varighet. / Filter by county, category, weather, season, indoor/outdoor, group size, price, and duration. Also supports near-me search via lat/lng (+ optional radius_km): when given, results include a rounded distance_km and a geo_precision flag ('address' = exact, 'kommune' = approximate municipality centroid) and are sorted nearest-first. Returns title, category, location (fylke/kommune), description, and booking URL if available. Only verified experiences from active providers (Brreg-checked) are returned. Covers Norway only: pass the user's request in query; for a place outside Norway (e.g. Kenya) the tool returns no experiences, and the answer is that Opplevagent only covers Norway. Food-product shopping such as buying stockfish is outside this tool's scope. Pass that intent in query; an out_of_scope response must not be retried as a general discovery search. Examples: 'hva kan vi finne på i Troms om vinteren?' (fylke='Troms', season='winter'), 'wildlife experiences' (category='dyreliv_safari'), 'outdoor activities in Oslo for 4 people', 'experiences within 50km of lat 69.65 / lng 18.95'.
| Name | Required | Description | Default |
|---|---|---|---|
| age | No | Age of the youngest participant. Filters out experiences with a minimum-age requirement above this. Example: 8 | |
| lat | No | Origin latitude for a near-me search (decimal degrees). Must be given together with lng. Example: 69.65 (Tromsø) | |
| lng | No | Origin longitude for a near-me search (decimal degrees). Must be given together with lat. Example: 18.95 (Tromsø) | |
| sort | No | 'distance' — sort results ascending by distance from lat/lng. Already the default whenever lat/lng are given; this makes the request explicit. | |
| fylke | No | Norwegian county (fylke). Examples: 'Oslo', 'Vestland', 'Troms', 'Rogaland' | |
| limit | No | Max results (default 20, max 50) | |
| query | No | The user's request in a few words, including any place they named — e.g. 'safari lodge in Kenya', 'northern lights in Tromsø'. Opplevagent only covers Norway: when this names a place outside Norway and no Norwegian fylke/kommune or coordinates are given, the tool returns no experiences and says so. Include the purchase intent too: buying stockfish is food retail, not an experience search. | |
| season | No | Season filter: 'summer', 'winter', 'spring' or 'autumn' (the Norwegian words sommer/vinter/vår/høst work too). Year-round experiences are always included. Use this (not category or weather) when the user asks what to do in a season, e.g. 'Troms om vinteren' → fylke='Troms', season='winter'. | |
| kommune | No | Norwegian municipality (kommune). Examples: 'Tromsø', 'Bergen', 'Stavanger' | |
| weather | No | Today's weather, not the season: 'rain'/'snow' prefers indoor + weather-independent experiences. Examples: 'rain', 'clear'. For 'in winter' use season instead. | |
| category | No | Experience category slug — one of: 'kultur_historie' (culture/history, museums), 'natur_friluft' (nature/outdoors), 'adrenalin_action', 'sightseeing_transport' (sightseeing, cruises, cable cars), 'overnatting_opplevelse' (stays), 'dyreliv_safari' (wildlife, whale/moose safaris, dog sledding), 'vinter_sno' (winter & snow), 'mat_drikke' (food & drink), 'velvaere_spa' (wellness/spa). Call list_experience_categories for live counts. Omit category for a general 'what can we do' question. 'In winter' / 'om vinteren' is a season, not a category: use season='winter' — 'vinter_sno' covers only snow activities and would hide e.g. northern-lights cruises. | |
| language | No | Required language for the experience. Examples: 'no', 'en' | |
| max_price | No | Maximum price per person in Norwegian kroner (NOK). Example: 500 | |
| radius_km | No | Max distance from lat/lng in kilometers. Only applies when lat/lng are given. Example: 50 | |
| group_size | No | Number of people in the group. Used to filter experiences by min/max group capacity. Example: 4 | |
| duration_max | No | Maximum duration in minutes. Example: 120 (2 hours) | |
| indoor_outdoor | No | Indoor/outdoor preference. Examples: 'indoor', 'outdoor', 'both' |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare read-only/idempotent/non-destructive and a closed world, so safety is covered. The description goes well beyond them: near-me results include a rounded distance_km and a geo_precision flag ('address' exact vs 'kommune' centroid), results are sorted nearest-first, only Brreg-verified providers with active status are returned, and out-of-scope requests produce a terminal out_of_scope response.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Front-loaded and well organized: scope, filters, geo behavior, return shape, then domain limits and examples. It is slightly wasteful in places — the filter list is duplicated in Norwegian and English, and the stockfish/out-of-scope rule is restated in the query parameter schema — but no sentence is filler.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a 17-parameter, zero-required tool with no output schema, the description supplies the return shape (title, category, fylke/kommune, description, booking URL), the domain boundary, the out-of-scope contract, and worked examples. An agent has everything needed to call it correctly on the first try.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100%, so the baseline is 3, and the description adds real value on top: it maps natural-language requests to parameter combinations (fylke='Troms', season='winter'), explains the season/weather/category distinction, and spells out the query field's dual role for place and purchase intent. It stops short of explaining a few fields (age, language, duration_max) beyond what the schema already says.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
States a specific verb (search) and resource (curated marketplace of Norwegian experiences) plus the exact filter axes it supports. It explicitly distinguishes itself from siblings by naming list_experience_categories for live category counts and by framing itself as the discovery entry point that get_experience complements.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Gives explicit routing rules: season vs weather vs category ('in winter' is a season, not a category), omit category for general 'what can we do' questions, use near-me only when lat/lng are supplied. It also states a hard boundary (Norway only) with the exact non-Norway behavior and warns that an out_of_scope response must not be retried as a general search.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Search opplevagent.no's gårdssalg (farm-sale) vertical — Brreg-registered Norwegian drink producers (bryggeri/cideri/vingård/destilleri/gårdskafé/mjød/seltzeri) selling directly from the farm. Finn norske gårdssalg-produsenter (drikkeprodusenter som selger direkte fra gården) etter fylke, kommune og produsenttype. / Filter by county (fylke), municipality (kommune), and producer type. Also supports near-me search via lat/lng (+ optional radius_km): when given, results include a rounded distance_km and are sorted nearest-first, and rows with no geocoded location are excluded (never a fabricated distance). Returns id (the provider_id for book_gardssalg), name, location, producer type, profile_url, and an honest booking status (live direct booking vs. a dark-launch 'coming soon' note — never overclaims booking availability). Include each returned producer's profile_url as a clickable link when available; when booking is paused, tell the user to contact the producer via that profile page. To look up ONE specific producer by name (e.g. when the guest says 'book hos Fjordgard Bryggeri'), pass query. Examples: 'gårdssalg i Vestland', 'cideri near lat 60.4 / lng 5.3', 'bryggeri med booking', query: 'Fjordgard Bryggeri'.
| Name | Required | Description | Default |
|---|---|---|---|
| lat | No | Origin latitude for a near-me search (decimal degrees). Must be given together with lng. Example: 69.65 (Tromsø) | |
| lng | No | Origin longitude for a near-me search (decimal degrees). Must be given together with lat. Example: 18.95 (Tromsø) | |
| fylke | No | Norwegian county (fylke) of the producer. Examples: 'Oslo', 'Vestland', 'Troms', 'Rogaland' | |
| limit | No | Max results (default 20, max 50) | |
| query | No | Free-text lookup of a specific producer by name and/or place, e.g. 'Fjordgard Bryggeri', 'Egge gård', 'sideri Hardanger'. Every word must match the producer's name, URL slug, place (poststed) or municipality; exact name matches rank first. Use this to get the `id` (provider_id) for book_gardssalg when the guest names a producer. A county name in the query (e.g. 'Vestland') is applied as the fylke filter; for a whole county, prefer fylke. | |
| kommune | No | Norwegian municipality (kommune) of the producer. Examples: 'Tromsø', 'Bergen', 'Stavanger' | |
| radius_km | No | Max distance from lat/lng in kilometers. Only applies when lat/lng are given. Example: 50 | |
| booking_live | No | When true, only return producers that currently accept direct bookings. Omit to include producers regardless of booking status. | |
| producer_type | No | Type of drink producer — one of the six canonical drink venue kinds: 'bryggeri' (brewery), 'cideri' (cidery), 'vingård' (winery), 'destilleri' (distillery), 'gårdskafé' (farm café), 'mjød'/'mjøderi' (meadery). 'seltzeri' also occurs in the data. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already cover the safety profile (readOnly, idempotent, non-destructive), and the description adds real behavioral detail on top: geocoded-location rows are excluded rather than distance-fabricated, distance_km is rounded and sorted nearest-first, and booking status is 'honest' (live vs. dark-launch 'coming soon') and never overclaimed. That is genuine disclosure beyond structured fields.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Purpose is front-loaded and the sentence order follows the agent's decision path (filter → near-me → query → outputs → examples). It loses a point for the Norwegian/English duplication of the same filtering sentence and for stacking output-field narration into a paragraph that could be tighter.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
With no output schema, the description carries the return contract and does so fully: id (cross-referenced to book_gardssalg), name, location, producer type, profile_url, and booking status. Combined with the examples and the near-me caveat, an agent has everything needed to call and interpret results correctly.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100%, so baseline is 3, but the description adds semantics the schema does not: query words must all match name/slug/poststed/municipality with exact-name matches ranked first, and a county name inside `query` is silently applied as the fylke filter. The lat/lng pairing and radius_km scoping are already in the schema, so it does not exceed a 4.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
States a precise verb+resource ('Search opplevagent.no's gårdssalg vertical — Brreg-registered Norwegian drink producers') and scopes it to direct-from-farm sellers across six named producer kinds. It also names the sibling it feeds into ('id — the provider_id for book_gardssalg'), so an agent can distinguish discovery from booking without opening either schema.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Explicit routing rules: use `query` for a single named producer, prefer `fylke` for whole-county lookups, use lat/lng for near-me search, and set `booking_live` to restrict to bookable producers. It also states the downstream action ('include profile_url as a clickable link', 'when booking is paused, tell the user to contact the producer'), which is unusually actionable guidance.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Fetch complete details for a single experience from opplevagent.no by its UUID. Henter fullstendig informasjon om en opplevelse via UUID. Returns title, full description, category, location (fylke/kommune/meeting point), indoor/outdoor, season, weather suitability, group size limits, age suitability, price, duration, languages, booking URL, and booking type. Obtain the UUID from discover_experiences results. Example: id 'a1b2c3d4-e5f6-7890-abcd-ef1234567890'.
| Name | Required | Description | Default |
|---|---|---|---|
| id | Yes | UUID of the experience to fetch. Example: 'a1b2c3d4-e5f6-7890-abcd-ef1234567890' |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare readOnlyHint=true, idempotentHint=true, and destructiveHint=false, covering the safety profile. The description adds valuable context beyond that: the exact source domain and a full inventory of returned fields (title, category, location, price, duration, booking URL, etc.). No contradiction with annotations.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The English and Norwegian sentences are near-duplicates, adding redundant length without new information. The returned-fields list and the UUID source guidance are useful, but the bilingual repetition could be trimmed to save tokens.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
With no output schema, the description compensates by enumerating all returned fields, which is essential for an agent to know what to expect. For a single-parameter, read-only tool with rich annotations, nothing critical is missing—minor gap is the absence of an explicit 'use when you have a UUID' trigger.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100% with format, pattern, and an example in the description. The description adds the critical workflow detail that the id comes from discover_experiences output, which is semantic value the schema alone does not convey.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description states a specific verb (Fetch), a precise resource (complete details for a single experience), and the source domain (opplevagent.no), with a brief Norwegian mirror for locality. It is clearly distinct from the sibling list/discovery tools, so an agent can tell this is the detail-retrieval counterpart to discover_experiences.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
It explicitly tells the agent where to obtain the required UUID ('Obtain the UUID from discover_experiences results'), establishing the correct workflow. It does not spell out when NOT to use it versus siblings, but the single-id resource makes the use case self-evident.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
List all experience categories available on opplevagent.no, along with the number of verified experiences in each category. Henter alle kategorier med antall verifiserte opplevelser. Use this to understand what kinds of experiences are available before calling discover_experiences with a specific category filter. Example question: 'hvilke typer opplevelser finnes i Norge?', 'what categories are available?'.
| Name | Required | Description | Default |
|---|---|---|---|
No parameters | |||
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare readOnlyHint=true and destructiveHint=false. The description adds that it returns counts of verified experiences, which is useful behavioral context beyond the safety profile.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Description is concise with two sentences in English and Norwegian, but the bilingual content is slightly redundant. Front-loaded with main purpose.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given no parameters, no output schema, and adequate annotations, the description provides sufficient information: what it does, when to use, and what it returns (categories with counts).
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
No parameters exist, so baseline is 4. No additional parameter meaning needed.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool lists all experience categories with counts, specifying the resource (categories on opplevagent.no) and action (list with number of verified experiences). This distinguishes it from siblings like discover_experiences.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Explicitly advises use before discover_experiences for category filter, with example questions like 'hvilke typer opplevelser finnes i Norge?'. This provides clear when-to-use context and differentiates from similar tools.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
discover_experiences1 field changedInput schema / properties / query / descriptionPrevious value: -"The user's request in a few words, including any place they named — e.g. 'safari lodge in Kenya', 'northern lights in Tromsø'. Opplevagent only covers Norway: when this names a place outside Norway and no Norwegian fylke/kommune or coordinates are given, the tool returns no experiences and says so."New value: +"The user's request in a few words, including any place they named — e.g. 'safari lodge in Kenya', 'northern lights in Tromsø'. Opplevagent only covers Norway: when this names a place outside Norway and no Norwegian fylke/kommune or coordinates are given, the tool returns no experiences and says so. Include the purchase intent too: buying stockfish is food retail, not an experience search."discover_experiences1 field changedInput schema / properties / queryAdded value: +{
+ "description": "The user's request in a few words, including any place they named — e.g. 'safari lodge in Kenya', 'northern lights in Tromsø'. Opplevagent only covers Norway: when this names a place outside Norway and no Norwegian fylke/kommune or coordinates are given, the tool returns no experiences and says so.",
+ "type": "string"
+}discover_experiences3 fields changedInput schema / properties / category / descriptionPrevious value: -"Experience category slug — one of: 'kultur_historie' (culture/history, museums), 'natur_friluft' (nature/outdoors), 'adrenalin_action', 'sightseeing_transport' (sightseeing, cruises, cable cars), 'overnatting_opplevelse' (stays), 'dyreliv_safari' (wildlife, whale/moose safaris, dog sledding), 'vinter_sno' (winter & snow), 'mat_drikke' (food & drink), 'velvaere_spa' (wellness/spa). Call list_experience_categories for live counts."New value: +"Experience category slug — one of: 'kultur_historie' (culture/history, museums), 'natur_friluft' (nature/outdoors), 'adrenalin_action', 'sightseeing_transport' (sightseeing, cruises, cable cars), 'overnatting_opplevelse' (stays), 'dyreliv_safari' (wildlife, whale/moose safaris, dog sledding), 'vinter_sno' (winter & snow), 'mat_drikke' (food & drink), 'velvaere_spa' (wellness/spa). Call list_experience_categories for live counts. Omit category for a general 'what can we do' question. 'In winter' / 'om vinteren' is a season, not a category: use season='winter' — 'vinter_sno' covers only snow activities and would hide e.g. northern-lights cruises."Input schema / properties / season / descriptionPrevious value: -"Season filter: 'summer', 'winter', 'spring' or 'autumn' (the Norwegian words sommer/vinter/vår/høst work too). Year-round experiences are always included."New value: +"Season filter: 'summer', 'winter', 'spring' or 'autumn' (the Norwegian words sommer/vinter/vår/høst work too). Year-round experiences are always included. Use this (not category or weather) when the user asks what to do in a season, e.g. 'Troms om vinteren' → fylke='Troms', season='winter'."Input schema / properties / weather / descriptionPrevious value: -"Weather suitability filter. 'rain'/'snow' prefers indoor + weather-independent experiences. Examples: 'rain', 'clear'"New value: +"Today's weather, not the season: 'rain'/'snow' prefers indoor + weather-independent experiences. Examples: 'rain', 'clear'. For 'in winter' use season instead."discover_gardssalg1 field changedInput schema / properties / query / descriptionPrevious value: -"Free-text lookup of a specific producer by name and/or place, e.g. 'Fjordgard Bryggeri', 'Egge gård', 'sideri Hardanger'. Every word must match the producer's name, URL slug, place (poststed) or municipality; exact name matches rank first. Use this to get the `id` (provider_id) for book_gardssalg when the guest names a producer."New value: +"Free-text lookup of a specific producer by name and/or place, e.g. 'Fjordgard Bryggeri', 'Egge gård', 'sideri Hardanger'. Every word must match the producer's name, URL slug, place (poststed) or municipality; exact name matches rank first. Use this to get the `id` (provider_id) for book_gardssalg when the guest names a producer. A county name in the query (e.g. 'Vestland') is applied as the fylke filter; for a whole county, prefer fylke."discover_experiences2 fields changedInput schema / properties / category / descriptionPrevious value: -"Experience category slug. Examples: 'natur_friluft', 'dyreliv_safari', 'mat_drikke', 'vinter'"New value: +"Experience category slug — one of: 'kultur_historie' (culture/history, museums), 'natur_friluft' (nature/outdoors), 'adrenalin_action', 'sightseeing_transport' (sightseeing, cruises, cable cars), 'overnatting_opplevelse' (stays), 'dyreliv_safari' (wildlife, whale/moose safaris, dog sledding), 'vinter_sno' (winter & snow), 'mat_drikke' (food & drink), 'velvaere_spa' (wellness/spa). Call list_experience_categories for live counts."Input schema / properties / season / descriptionPrevious value: -"Season filter. Examples: 'summer', 'winter', 'spring', 'autumn'"New value: +"Season filter: 'summer', 'winter', 'spring' or 'autumn' (the Norwegian words sommer/vinter/vår/høst work too). Year-round experiences are always included."discover_gardssalg1 field changedInput schema / properties / producer_type / descriptionPrevious value: -"Type of drink producer. Examples: 'bryggeri' (brewery), 'cideri' (cidery), 'vingård' (vineyard), 'destilleri' (distillery), 'mjøderi' (meadery), 'seltzeri'"New value: +"Type of drink producer — one of the six canonical drink venue kinds: 'bryggeri' (brewery), 'cideri' (cidery), 'vingård' (winery), 'destilleri' (distillery), 'gårdskafé' (farm café), 'mjød'/'mjøderi' (meadery). 'seltzeri' also occurs in the data."book_gardssalg4 fields changedInput schema / properties / provider_id / descriptionPrevious value: -"The gårdssalg producer's id, from a discover_gardssalg result (NOT the profile slug). Example: '3f1b2c4d-...'"New value: +"The gårdssalg producer's id — the `id` field of a discover_gardssalg result (NOT the profile slug). Example: '3f1b2c4d-...'. Either provider_id or provider_query is required; provider_id wins when both are given."Input schema / properties / provider_queryAdded value: +{
+ "description": "Alternative to provider_id: the producer's name as the guest said it, optionally with a place, e.g. 'Fjordgard Bryggeri' or 'Egge gård Steinkjer'. Resolved to exactly ONE producer; if several match, the tool returns the candidates (no booking created) so you can ask the guest which one; if none match, it says so.",
+ "maxLength": 200,
+ "type": "string"
+}Input schema / properties / requested_weekdayAdded value: +{
+ "description": "Safety check — the weekday the guest actually said, e.g. 'fredag' or 'Friday'. If slot_at does not fall on that weekday (Europe/Oslo) the tool returns weekday_mismatch:true with the nearest dates that do, instead of booking the wrong day. Always pass it when the guest named a weekday.",
+ "maxLength": 20,
+ "type": "string"
+}Input schema / requiredPrevious value: -[
- "provider_id",
- "slot_at",
- "party_size",
- "guest_name",
- "guest_email"
-]New value: +[
+ "slot_at",
+ "party_size",
+ "guest_name",
+ "guest_email"
+]discover_gardssalg1 field changedInput schema / properties / queryAdded value: +{
+ "description": "Free-text lookup of a specific producer by name and/or place, e.g. 'Fjordgard Bryggeri', 'Egge gård', 'sideri Hardanger'. Every word must match the producer's name, URL slug, place (poststed) or municipality; exact name matches rank first. Use this to get the `id` (provider_id) for book_gardssalg when the guest names a producer.",
+ "maxLength": 200,
+ "type": "string"
+}book_gardssalg1 field changedInput schema / properties / confirm_outside_hoursAdded value: +{
+ "description": "Set to true only when RETRYING after a previous call to this tool returned outside_hours:true and the requested time should be kept anyway, despite falling outside the producer's stated opening hours. Omit on a first attempt.",
+ "type": "boolean"
+}book_gardssalgdiscover_gardssalgdiscover_experiences4 fields changedInput schema / properties / latAdded value: +{
+ "description": "Origin latitude for a near-me search (decimal degrees). Must be given together with lng. Example: 69.65 (Tromsø)",
+ "maximum": 90,
+ "minimum": -90,
+ "type": "number"
+}Input schema / properties / lngAdded value: +{
+ "description": "Origin longitude for a near-me search (decimal degrees). Must be given together with lat. Example: 18.95 (Tromsø)",
+ "maximum": 180,
+ "minimum": -180,
+ "type": "number"
+}Input schema / properties / radius_kmAdded value: +{
+ "description": "Max distance from lat/lng in kilometers. Only applies when lat/lng are given. Example: 50",
+ "exclusiveMinimum": 0,
+ "maximum": 5000,
+ "type": "number"
+}Input schema / properties / sortAdded value: +{
+ "description": "'distance' — sort results ascending by distance from lat/lng. Already the default whenever lat/lng are given; this makes the request explicit.",
+ "enum": [
+ "distance"
+ ],
+ "type": "string"
+}discover_experiencesget_experiencelist_experience_categorieslokal_discoverlokal_get_producer_affiliationslokal_get_umbrella_memberslokal_infolokal_list_umbrellaslokal_searchlokal_statsdiscover_experiencesget_experiencelist_experience_categorieslokal_discoverlokal_get_producer_affiliationslokal_get_umbrella_memberslokal_infolokal_list_umbrellaslokal_searchlokal_statsdiscover_experiencesget_experiencelist_experience_categoriesNorwegian local-food MCP: 1,500+ farms, shops, markets + umbrellas (Bondens marked, Hanen, REKO).
Entur MCP — Norway public transport, nationwide, all modes (developer.entur.org)
Stortinget (Norwegian Parliament) open data MCP — data.stortinget.no.
Norwegian company search, accounts, roles, groups and websites from Brønnøysund, remote MCP.
Add one secure layer between your agents and this server.