The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references an installation script from packagecloud.io, a well-known service for hosting package repositories, to provide the official installation method for CrowdSec. Evidence: curl -s https://packagecloud.io/install/repositories/crowdsec/crowdsec/script.deb.sh | sudo bash in SKILL.md.\n- [COMMAND_EXECUTION]: Multiple administrative commands are documented for managing the security agent, viewing alerts, and modifying firewall decisions using sudo. Evidence: sudo cscli decisions add, sudo cscli alerts list, and sudo systemctl reload crowdsec across SKILL.md and references/configuration.md.\n- [PROMPT_INJECTION]: The skill documents the ingestion of external log sources via acquis.yaml, creating a potential surface for indirect prompt injection through log manipulation.\n
Ingestion points: Log filenames and journald filters defined in references/configuration.md.\n
Boundary markers: No explicit markers or delimiters are described for separating untrusted log content from agent instructions.\n
Capability inventory: The skill documents the capability to perform system-level actions such as banning IPs or reloading services via sudo cscli and systemctl.\n
Sanitization: Relies on the CrowdSec agent's internal parser logic to validate and normalize external data.

crowdsec