The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill uses the requests library in scripts/tempest-cli to fetch weather data and forecasts from the official WeatherFlow REST API at swd.weatherflow.com.
[COMMAND_EXECUTION]: The skill includes a standalone Python script scripts/tempest-cli which is designed to be executed by the agent to perform API queries and listen for local network broadcasts.
[DATA_EXFILTRATION]: The skill requires a user-provided API token (TEMPEST_TOKEN), which it transmits to swd.weatherflow.com via query parameters. This is the intended authentication mechanism for the service.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection (Category 8) because it processes untrusted data from external sources.
Ingestion points: Data is ingested via REST API calls to swd.weatherflow.com and local UDP broadcasts on port 50222 in scripts/tempest-cli.
Boundary markers: The script does not implement specific delimiters or warnings to the agent context to ignore instructions embedded within the weather data.
Capability inventory: The script scripts/tempest-cli does not possess dangerous capabilities; it lacks file-writing, arbitrary command execution (eval/exec/subprocess), or complex logic that could be subverted by malicious input. Its primary function is data retrieval and formatting.
Sanitization: The script performs basic sanitization by casting positional array values to specific types (float, int) before formatting them into strings.
[DATA_EXPOSURE]: The script scripts/tempest-cli attempts to read a configuration file from the user's home directory at ~/.tempest.env. This is a standard pattern for CLI credential management and does not represent an unusual exposure.

tempest-cli