The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill repeatedly uses the pattern of piping curl output directly into python3 to process JSON data. While some instances use the -c flag with inline code, this pattern creates a high-risk execution vector where a compromised service (Prometheus on localhost or the Telegram API) could potentially trigger arbitrary code execution on the host system.
Evidence in scripts/stack-health.sh: curl -sf "https://api.telegram.org/bot${BOT_TOKEN}/getMe" 2>/dev/null | python3 -c "..."
Evidence in references/alert-rules-guide.md: curl -s http://localhost:9472/api/v1/rules | python3 -c "..."
[CREDENTIALS_UNSAFE]: The skill contains logic in scripts/stack-health.sh to programmatically search for and extract the Telegram bot_token from a configuration file (~/docker/stacks/monitoring/alertmanager/config.yml) using grep and awk. This facilitates the exposure of sensitive credentials to the agent's context.
[COMMAND_EXECUTION]: The skill utilizes several shell commands for stack management, including docker compose, ps aux, and sensors. While these are functional for monitoring, they represent a significant capability surface when combined with the data processing patterns found in the scripts.
[EXTERNAL_DOWNLOADS]: The skill makes network requests to api.telegram.org to verify bot status and test alert delivery. While Telegram is a well-known service, these requests are used as part of the execution pipe to Python mentioned above.

monitoring-stack