mikahama/uralicnlp/uralicnlp-skill-template

__LANG_NAME__ morphology

Quick start

1) Ensure dependencies are installed

If imports fail or this is the first run in the session, install UralicNLP:

python -m pip install -r scripts/requirements.txt

2) Use the CLI helper for deterministic results

All commands output JSON.

• Morphological analysis:

python scripts/uralic_cli.py analyze --word mieʹcc

• Lemmatize:

python scripts/uralic_cli.py lemmatize --word mieʹcen

• Generate/inflect from a full analysis string:

python scripts/uralic_cli.py generate --inflection mieʹcc+N+Sg+Gen

• Translate a lemma:

python scripts/uralic_cli.py translate --lemma mieʹcc

How to respond to users

Inputs to request (only when missing)

• word or inflection string
• lemma for translations
• The language is fixed to __LANG_NAME__ (__LANG_ISO__); do not ask the user to choose another language.

Output conventions

• Prefer returning:
• analysis as a list of strings like mieʹcc+N+Sg+Nom
• lemmatization as a list of lemmas
• translation as the JSON value returned by the translation lookup
• generation as a list of surface forms
• If the CLI returns { "error": ... }, explain what went wrong and suggest the next action, usually installing deps or checking that the bundled HFST-OL files are present.
• This skill works offline by using the bundled HFST-OL files in scripts/. Do not rely on UralicNLP downloading models.

Script reference

• scripts/uralic_cli.py: main entrypoint. Use it instead of rewriting code in-chat.
• scripts/analyser-gt-desc.hfstol: bundled __LANG_NAME__ analyzer model.
• scripts/dict1.hfstol: first bundled __LANG_NAME__ translation dictionary.
• scripts/dict2.hfstol: second bundled __LANG_NAME__ translation dictionary.
• scripts/generator-gt-norm.hfstol: bundled __LANG_NAME__ generator model.
• scripts/requirements.txt: dependency list.