Default Tools

Chalie ships with a curated set of default tools that are installed automatically on first startup. These tools are listed in backend/configs/embodiment_library.json with "installs_by_default": true.

Default tools are trusted — they run as subprocesses in Chalie’s Python environment rather than inside Docker containers. This means they start instantly and don’t require Docker to be installed.

Auto-Install Behavior

On first startup, if a default tool is not present in backend/tools/, Chalie fetches the latest release tarball from the tool’s GitHub repository and installs it automatically (background thread, non-blocking). Subsequent startups skip tools that are already present.

To opt out of this behavior at install time:

curl -fsSL https://chalie.ai/install | bash -s -- --disable-default-tools

This writes a backend/data/.no-default-tools marker file. Chalie will not auto-install default tools as long as this file exists.

Note: Disabling default tools does not prevent you from installing them manually later via the Tools → Catalog UI or the POST /api/tools/install endpoint.

Installed by Default

Weather


Repo	chalie-ai/chalie-tool-weather
Category	Context
Trigger	On-demand
Trust	Trusted (subprocess)
Requires API key	No

Fetches current weather conditions and tomorrow’s forecast using Open-Meteo (coordinates-based, primary) and wttr.in (city name fallback). Both sources are free with no API key required.

When Chalie uses it: Any time weather context is relevant — explicit questions (“what’s the weather like?”), implied context (“should I bring a jacket?”), or when outdoor conditions matter to an activity (“I’m going for a run”).

Returns:

Field	Description
`temperature_c` / `temperature_f`	Current temperature
`feels_like_c`	Apparent temperature
`condition`	Human-readable description (e.g. “Partly cloudy”)
`humidity_pct`	Relative humidity %
`wind_kmh` / `wind_direction`	Wind speed and compass direction
`visibility_km`	Visibility
`uv_index`	UV index
`precip_mm`	Precipitation
`is_raining` / `is_hot` / `is_cold` / `is_windy` / `is_clear` / `is_daylight`	Boolean condition flags
`forecast_tomorrow_condition`	Tomorrow’s condition description
`forecast_tomorrow_max_c` / `_min_c`	Tomorrow’s temperature range
`forecast_tomorrow_precip_chance_pct`	Tomorrow’s rain probability

Results are cached per location for 10 minutes. Location is detected automatically from client telemetry — only pass an explicit location parameter when asking about a different place.

Adding More Default Tools

To mark a new tool as a default, add an entry to backend/configs/embodiment_library.json:

{
  "name": "my_tool",
  "title": "My Tool",
  "icon": "fa-star",
  "repo": "chalie-ai/chalie-tool-my-tool",
  "summary": "One-line description shown in the catalog",
  "category": "utility",
  "trigger": "on_demand",
  "trust": "trusted",
  "installs_by_default": true
}

Requirements for trusted default tools:

Must have a runner.py (subprocess entry point) — no Dockerfile needed
Must have a manifest.json with valid name, description, trigger, parameters, returns
Must be hosted on GitHub with at least one tagged release
Must not bundle any secrets or environment-specific configuration