Ski Assistant

Name/description match the packaged assets: a resorts DB, search strategies, modules for pricing, coaching, discovery, and five Python tools that implement network lookups and card generation. Requested resources (local DB, webfetch, websearch, optional flyai) are appropriate for trip planning, pricing and coaching.

SKILL.md explicitly instructs the agent to read/write files in ~/.ski-assistant (user_profile.json, records.json, watchlist.json, upcoming_trips.json, price_history.json, custom_resorts.json). It also instructs calling Python tools and external APIs (Open-Meteo, exchangerate-api, Overpass/Nominatim, raw.githubusercontent.com, websearch). These operations are coherent with the described features but are material: installing will give the skill local storage and network access. The README/skill require the agent to run python3 scripts, but the skill metadata lists no required binaries — an inconsistency to note.

There is no install spec (instruction-only skill with bundled code). The included Python files are executed via Bash(python3 ...) per SKILL.md; nothing is downloaded from arbitrary hosts at install time. The database update feature can pull from raw.githubusercontent.com when the update-db tool is invoked, which is a reasonable, documented data-sync behavior.

The skill requests no environment variables or secret credentials beyond an optional SKI_ASSISTANT_DATA_DIR for changing the data directory. That matches its stated functionality. It does not require unrelated cloud keys or tokens.

The skill persists user data under ~/.ski-assistant and may create scheduled reminders (cron) only with explicit user consent and platform support. Persistent storage and optional cron tasks are within the skill's scope (saving coaching records, watchlists, upcoming trips), but users should be aware these files are written to their home directory and that creating system cron jobs (if allowed) can have broader effects.