Get started
AI & agents

Agent skills

Agent skills

Teach your coding agent to build with Mountaya. One command installs seven skills into Claude Code, Cursor, Codex, or any agent that reads Agent Skills:

Terminal
npx skills add https://docs.mountaya.com

Your agent loads a skill only when a task needs it. It then writes Mountaya code with the right keys, hosts, and limits.

The skills

SkillTeaches your agent to
mountaya-authPick keys and scopes, mint session tokens, retry writes safely, and read errors and rate limits
mountaya-data-apiRoute, geocode, search, export itineraries, and read terrain with the Data API
mountaya-tile-apiAdd Mountaya base maps and overlays to MapLibre, Mapbox GL JS, or deck.gl
mountaya-map-publishingEmbed published maps with the iframe, its parameters, and its events
mountaya-webhooksReceive webhook events: verify signatures, handle retries and duplicates, and recover missed events
mountaya-sdk-typescriptChoose and use the @mountaya/* packages, from the API client to Embedded Studio
mountaya-mcpConnect assistants to the MCP server, and pick its tools and prompts

The other six skills rely on mountaya-auth for keys, tokens, and errors.

Install some of them

Pass --skill once for each skill you want:

Terminal
npx skills add https://docs.mountaya.com --skill mountaya-auth --skill mountaya-data-api

Installed skills don't update themselves. Run npx skills update -y to get new versions.

Set your keys

The Data API skill runs scripts that read your keys from the environment:

Terminal
export MOUNTAYA_PUBLISHABLE_KEY="pk_your_publishable_key"
export MOUNTAYA_SECRET_KEY="sk_your_secret_key"

Give the publishable key the scopes your work needs, such as data for the Data API. Create both keys under Organization → API keys in the Mountaya app.

Where they live

The skills are maintained in the public mountayaapp/ai repository and served from this site. An assistant connected to the MCP server gets the same procedures as prompts, with nothing to install.

Next steps

Last updated