The Agent Skills Directory

[COMMAND_EXECUTION]: The file scripts/sentinel_config.py contains a utility function mcporter_call that utilizes subprocess.run to execute the mcporter CLI. This is the standard, documented method for skills on this platform to interact with the trading backend. The function is used programmatically with hardcoded tool identifiers, which mitigates the risk of command injection from external data.- [SAFE]: The skill implements best practices for secret management by utilizing environment variables (e.g., MANTIS_WALLET) and local configuration files (config/sentinel-config.json) with placeholders, ensuring that sensitive credentials are not hardcoded.- [SAFE]: All external market data retrieved via API calls is processed using deterministic mathematical logic and scoring algorithms. The skill does not pass untrusted content back to the LLM for interpretation, effectively neutralizing the risk of indirect prompt injection.

sentinel-strategy