Parking Radar

Find, verify, and reserve parking worldwide with provider selection, live-signal triage, and local memory for favorite places and cities.

When to Use

User needs parking now, wants to pre-book for an airport or event, needs the right local parking app for a city, or wants help comparing parking options across providers.

Use this skill when the agent should distinguish maps search from real reservations, choose the right operator or marketplace by country and city, track favorite parking spots, and remember hard-won parking discoveries for later.

Architecture

Memory lives in ~/parking/. If ~/parking/ does not exist, run setup.md. See memory-template.md for structure.

~/parking/
├── memory.md          # Activation defaults, home base, vehicle profile, and booking boundaries
├── favorites.md       # User-approved recurring car parks, venues, and trusted entrances
├── cities.md          # City-by-city provider choices, fallback order, and local quirks
├── findings.md        # User-confirmed discoveries, restrictions, scams, and quality signals
├── sessions/
│   └── YYYY-MM-DD.md  # Complex search trail when a trip or event needs multiple comparisons
└── archive/           # Older notes that should not stay in the hot path

Quick Reference

Load only the smallest file needed for the current parking job.

Topic	File
Setup and activation behavior	setup.md
Memory schema and local note templates	memory-template.md
Global provider and market matrix	provider-registry.md
Major city patterns and fallback order	city-patterns.md
Short workflows for search, booking, and verification	execution-playbook.md
Public and partner API notes	api-notes.md
Official research source ledger	source-ledger.md

Requirements

• No credentials are required for planning, provider selection, or official web research.
• Live bookings, plate entry, or payment flows require the user to approve the chosen provider and any data sent to it.
• Ask before creating persistent local parking memory. Offer a one-off mode when the user does not want favorites, city defaults, or discovery notes saved.
• Do not promise a guaranteed space unless the provider explicitly exposes reservation inventory or confirmed booking status.

Coverage

This skill is built for the parking cases where generic map search is not enough:

• quick parking nearby right now
• airport, train-station, stadium, and venue parking that benefits from pre-booking
• repeated home-city parking where favorites, vehicle limits, and local operator habits matter
• travel planning where the correct provider changes by country or even by city
• parking platform research and API scouting for product or integration work

Core Rules

1. Classify the parking mode before searching

• Separate the job into one of five modes: find-now, pre-book, recurring-city, operator research, or API scouting.
• Parking answers get worse fast when the agent mixes instant parking, on-street payment, and guaranteed reservation into one workflow.

2. Default to the lightest successful path

• If the user just needs parking now, start with map discovery plus the strongest official operator or aggregator for that city.
• Use deeper provider comparison only when the quick path is ambiguous, expensive, full, or clearly unreliable.
• Do not turn a simple "find me parking nearby" request into a long market audit.

3. Distinguish discovery, payment, and reservation surfaces

• Google Maps, Apple Maps, and Parkopedia are discovery surfaces first.
• Apps such as RingGo, PayByPhone, EasyPark, APCOA Connect, and Parking.sg often solve payment or session start, not guaranteed inventory.
• Reservation marketplaces such as SpotHero, ParkWhiz, JustPark EventPass, Onepark, Parclick, Q-Park Prebook, and many operator-native airport or venue flows are the right surfaces for guaranteed entry when available.

4. Verify reservation-critical constraints before recommending

• Check entry window, exit rules, in-and-out privileges, height limit, EV charging, motorcycle or van support, license-plate requirements, and cancellation terms.
• For airports and events, confirm walking distance, shuttle dependency, venue match, and whether the booking is for a specific car park or only a zone.
• A cheap parking option is wrong if the user's vehicle or timing cannot actually use it.

5. Be precise about availability language

• "Reservable now" means the provider is offering a booking path for the requested window.
• "Reported open" means a listing says the facility is operating, but not that a bay is guaranteed.
• "Real-time occupancy" should only be claimed when an official provider or city API exposes it directly.

6. Keep a local parking memory only with consent

• Offer to remember home city, recurring venues, favorite facilities, vehicle constraints, and trusted providers locally in ~/parking/.
• Store durable notes, not raw payment data or full booking receipts.
• Promote user-confirmed discoveries such as hidden entrances, repeat scams, and reliable operator patterns into the local favorites, cities, or findings notes.

7. Treat city and country as first-class routing signals

• Use provider-registry.md to choose the right platform family for the market before assuming Google results or one familiar app will work.
• Dense urban centers often use payment apps and operator-native garages, while airports and major venues often use reservation marketplaces.
• When entering a new market, verify the dominant local stack instead of assuming North American or UK patterns apply worldwide.

8. Keep research evidence attached to local discoveries

• Save only discoveries that came from the user, an official provider, or a clearly identified city or operator source.
• Tag low-confidence hearsay as unverified and keep it out of favorites until it repeats or gets confirmed.
• Use source-ledger.md and api-notes.md to separate facts, partnership-only APIs, and "no public API found" cases.

Common Traps

• Treating a pay-by-phone parking app as proof that a space can be reserved -> many apps start a session only after the driver finds a bay.
• Claiming "spots available" from a static listing page -> open hours and active inventory are not the same thing.
• Ignoring plate, height, or vehicle-type restrictions -> the booking looks fine until gate entry fails.
• Recommending street parking for a high-stress event or airport drop -> uncertainty and enforcement risk overwhelm the small savings.
• Assuming one global app dominates every market -> parking is highly local and often city-specific.
• Saving raw booking confirmations or payment details -> keep reusable parking knowledge, not sensitive receipts.

External Endpoints

Endpoint	Data Sent	Purpose
https://maps.google.com and https://maps.apple.com	place text, coordinates, route context	Quick parking discovery, map links, and local navigation handoff
https://www.parkopedia.com and Parkopedia parking data products	location, time window, optional vehicle filters	Global parking discovery, facility detail, and broad market coverage
https://spothero.com and https://api.parkwhiz.com	destination, venue, dates, time window, optional vehicle metadata	North America reservation search and parking booking APIs
https://www.justpark.com, https://eventpass.justpark.com, https://ringgo.co.uk, https://www.q-park.co.uk, https://www.apcoaconnect.com, https://www.paybyphone.com	location code, destination, plate, and time window	UK and Europe parking reservation or payment workflows
https://parclick.com, https://www.onepark.co, https://parkmobile.io, https://parkplus.io, https://parking.sg, and official city data portals such as https://api.data.gov.hk	destination, date range, vehicle type, plate when required, and occupancy query parameters	Regional booking, payment, or official parking data lookups

No other data should be sent externally unless the user approves another official provider or city source.

Security & Privacy

Data that may leave your machine:

• parking destination or venue
• city and time window
• coordinates and route context
• plate number only when the chosen provider requires it for entry or billing
• vehicle profile fields such as height or EV requirement when they change the result

Data that stays local:

• favorite parking places and entrances in the local favorites note
• home-city defaults and provider order in the local city notes
• user-confirmed discoveries and warnings in the local findings note
• activation rules and parking boundaries in ~/parking/memory.md

This skill does NOT:

• store payment card data, account passwords, or unnecessary booking receipts
• claim guaranteed inventory without provider evidence
• use undeclared anti-bot scraping flows
• modify its own SKILL.md

Trust

This skill can send parking queries, dates, route context, plate numbers, or vehicle constraints to the selected parking marketplace, operator, payment app, maps provider, or official city data source. Only use live parking flows if you trust the chosen provider with that data.

Scope

This skill ONLY:

• finds, compares, and explains parking options
• chooses the right provider family for each city and task
• verifies reservation-critical constraints before the user commits
• maintains lightweight local memory for favorites, cities, and discoveries after approval

This skill NEVER:

• invent live availability or occupancy
• assume a payment app guarantees entry
• treat one country's parking workflow as universal
• persist sensitive payment or identity data without a clear need

Related Skills

Install with clawhub install <slug> if user confirms:

• maps - Use map providers, geocoding, and route links to ground parking search in real geography.
• travel - Fold parking choices into broader trip timing, lodging, and movement decisions.
• booking - Compare reservation terms, cancellation rules, and booking friction across providers.
• compare - Turn multiple parking options into a clear side-by-side decision.
• car-rental - Add pickup, return, and vehicle-access context when parking decisions sit inside a driving trip.

Feedback

• If useful: clawhub star parking
• Stay updated: clawhub sync