# AI resources for Statistics Greenland

This directory contains public, LLM-friendly guidance for agents using Statistics Greenland data.

## Recommended composition

1. Start with the compact Statbank quickstart:
   https://stat.gl/ai/statbank-quickstart.md
2. Use the canonical generic PxWeb API v1 skill for discovery, metadata and raw API querying:
   https://github.com/janbrus/pxwebapi-skills/tree/main/pxwebapi-v1-generic-skill
3. Apply the local Statistics Greenland profile for language-specific source links and provenance:
   https://stat.gl/ai/skills/statistics-greenland-statbank/SKILL.md
4. When R is available, prefer the Statistics Greenland `statgl` skill for Statbank retrieval and presentation workflows:
   https://stat.gl/ai/skills/statgl/SKILL.md
5. Use the `pxmake` skill when working with pxmake-specific R workflows.

The generic PxWeb API v1 skill is intentionally referenced upstream rather than mirrored here, so agents can discover the current canonical version instead of a stale local copy.

The rOpenGov PxWeb API catalogue is a useful discovery seed but is not exhaustive. Absence from the catalogue does not imply that a PxWeb API v1 installation is unsupported.

For Statistics Greenland itself, table discovery can be performed directly with HTTP GET `?query=` on the PxWeb API, for example `https://bank.stat.gl/api/v1/en/Greenland?query=population` or under a subject path such as `https://bank.stat.gl/api/v1/en/Greenland/BE?query=population`. The R equivalent is `statgl::statgl_search()`.

## Discovery

- Root agent entry point: https://stat.gl/llms.txt
- Machine-readable skill list: https://stat.gl/ai/skills.json
- Statbank-specific entry point: https://bank.stat.gl/llms.txt
- Recommended web discovery relation: `rel="describedby"` pointing to the applicable `llms.txt` file.

The deployment examples in the package are recommendations and should be merged into existing IIS/robots/sitemap configuration rather than copied over blindly.