AI Agents
Mountaya provides agent skills: structured reference files that teach AI coding agents how to integrate with Mountaya APIs. Install them once, and your agent handles authentication, embedding, geospatial queries, and tile rendering on your behalf.
The skills live in the open-source mountayaapp/agent-skills repository.
Skills
mountaya-authis the prerequisite for the other skills. It sets up publishable keys, secret keys, session token flows, and rate limits without you having to read the docs yourself.mountaya-embeddingpowers Map Embedding. It generates<iframe>snippets with the right URL parameters, postMessage handlers, and safety rules.mountaya-data-apipowers Outdoor Intelligence. It builds GraphQL queries for directions, route suggestions, travel-time matrices, isochrones, and geometry analysis. Depends onmountaya-authfor session tokens.mountaya-tiles-apipowers Outdoor Tiles. It adds weather overlays, routes, collections, and terrain layers to a MapLibre GL JS map.
Set up
1. Configure environment variables
The skills read your API credentials from environment variables, so your agent never sees them as plain text. Set them before launching your coding tool:
MOUNTAYA_PUBLISHABLE_KEYis required for every skill. It identifies your organization in every request.MOUNTAYA_SECRET_KEYis required formountaya-authandmountaya-data-api. It mints short-lived session tokens server-side. Keep it out of client code.
Create both keys from your organization's API settings. Enable the scopes your products need: data for Outdoor Intelligence, tiles for Outdoor Tiles, embedding plus tiles for Map Embedding. For the full authentication model, see Authentication.
2. Install the skills
Install all skills at once:
If you install mountaya-data-api on its own, install mountaya-auth alongside it. mountaya-data-api relies on it for session-token creation.
Manual installation
How agents use the skills
Once installed, skills are loaded automatically. When you ask your agent to work on a Mountaya integration (setting up authentication, embedding a map, querying Outdoor Intelligence, or adding tile layers) it picks up the relevant skill and follows the documented patterns.
No manual invocation is needed. Just describe what you want to build and the agent uses the skills as context.