The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/review.py script utilizes subprocess.run to call the openclaw CLI and curl. Analysis confirms these calls use argument lists rather than shell strings, which mitigates command injection risks.\n- [EXTERNAL_DOWNLOADS]: The skill communicates with the vendor's MCP server at mcp.prod.senpi.ai and the Hyperliquid API at api.hyperliquid.xyz. These are categorized as authorized vendor resources and well-known technology services respectively.\n- [CREDENTIALS_UNSAFE]: Sensitive authentication is managed through the SENPI_AUTH_TOKEN environment variable. This aligns with industry standards for secure secret management and avoids hardcoding credentials.\n- [DYNAMIC_EXECUTION]: The scripts/_yaml.py utility implements a custom YAML loader for environment compatibility, which correctly falls back to yaml.safe_load when the PyYAML library is present, ensuring secure data parsing.\n- [DATA_EXPOSURE]: The skill processes trade history and telemetry data to generate coaching narratives. This activity is restricted to authorized platform tools and is necessary for the skill's primary stated purpose.

senpi-improve-trades