The Agent Skills Directory

[PERSISTENCE]: The skill provides instructions to modify the system's cron table using (crontab -l 2>/dev/null; echo '0 7 * * * ~/.config/moonpay/scripts/messari-alpha-scout.sh # messari:alpha') | crontab -. This configuration ensures that a generated script is executed automatically every morning, establishing a persistence mechanism on the host system.
[INDIRECT_PROMPT_INJECTION]: The workflow retrieves trending topics, news, and market signals from the external Messari API (api.messari.io) and passes this content directly into an AI narrative synthesis prompt in Step 5.
Ingestion points: Data is fetched in Steps 1 through 4 and stored in local JSON files before being interpolated into the AI prompt.
Boundary markers: The AI prompt uses JSON interpolation ({step1_output}, etc.) without explicit boundary markers or instructions to ignore embedded commands within the fetched content.
Capability inventory: The agent environment includes capabilities for file system operations, network requests via mp x402, and financial transactions via mp token swap.
Sanitization: There is no evidence of sanitization or validation of the news headlines or mindshare data before they are processed by the LLM.
[COMMAND_EXECUTION]: The skill generates a local shell script at ~/.config/moonpay/scripts/messari-alpha-scout.sh by combining multiple CLI commands. This script is then executed by the system scheduler. While these commands use the vendor's own mp CLI and target a well-known service (Messari), the automated execution of generated scripts presents a broader attack surface if the script contents were to be influenced by external inputs.

messari-alpha-scout