avlihachev/mening-skill

Mening — daily language writing practice

Mening gives the user one writing prompt a day in the language they are learning, corrects what they write, and remembers the mistakes they repeat over time — that long-term error-pattern memory is the point, not one-off corrections.

The user's target language, level, and history live in their Mening account. This skill talks to that account over its API. The token is in MENING_API_TOKEN.

All commands go through the helper in this skill's folder: bash scripts/mening.sh <command>. Each prints JSON; read it and reply conversationally.

When to use

• "give me today's topic" / "what should I write about" → bash scripts/mening.sh today and read out the prompt text from prompts[].
• The user writes a sentence in their target language → bash scripts/mening.sh submit "<their exact sentence>". The response has the corrected text plus the specific mistakes found. Relay the correction, then say which recurring patterns it touched if any.
• "how am I doing" / "what do I keep getting wrong" → bash scripts/mening.sh trends and summarise patterns[] (each is a recurring error category + tag) and the streak in activity.
• "give me another topic" → bash scripts/mening.sh prompt (capped a few per day; if it returns a limit message, tell the user to come back tomorrow).

Rules

• Submit the user's sentence verbatim — do not pre-correct or translate it; the correction is Mening's job.
• There is a daily limit (~10 corrections). If a call returns an error about a limit, relay it kindly; do not retry.
• Never invent a correction or a pattern — only report what the API returns.
• If MENING_API_TOKEN is missing or a call returns "invalid or expired token", tell the user to open Settings → Developer / API access at https://mening.app and paste a fresh token.