tempest
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill follows security best practices for handling API tokens by using environment variables and local configuration files (~/.tempest.env) rather than hardcoding credentials.
- [SAFE]: All external network communication is directed to official WeatherFlow domains (swd.weatherflow.com) or the local network for UDP broadcasts (port 50222) as described in the documentation.
- [INDIRECT_PROMPT_INJECTION]: The skill processes weather data from external sources (WeatherFlow REST API and local UDP broadcasts). While this represents an attack surface for indirect prompt injection if the source data were compromised, the data is parsed as structured JSON/arrays and not executed.
- Ingestion points:
scripts/tempestviaTempestClient._get(REST API) andudp_listen(UDP datagrams). - Boundary markers: Absent. The skill output is delivered directly to the agent context without specialized delimiters.
- Capability inventory: The script performs network GET requests, local socket binding for UDP listening, and reads a specific local environment file.
- Sanitization: Data is strictly parsed using standard JSON decoders and mapped to predefined positional layouts before being returned to the agent.
Audit Metadata