The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill instructions facilitate the retrieval of administrative credentials by reading the root API key from /opt/ludus/install/root-api-key.
[COMMAND_EXECUTION]: The troubleshooting guide includes several high-risk commands and scripts:
Destructive file operations including recursive deletion (rm -rf /opt/ludus) and automated deletion via find with the -delete flag.
System-level configuration changes involving systemctl, iptables, ifup, and update-ca-certificates.
Execution within a modified root environment via chroot for GRUB repair procedures.
[EXTERNAL_DOWNLOADS]: The skill references and instructs the user to download assets from the vendor's GitLab repository (gitlab.com/badsectorlabs/ludus.git) and official documentation domains (docs.ludus.cloud).
[PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted data sources which could lead to command manipulation.
Ingestion points: Analyzes deployment logs, error messages, and range configurations provided by the user or system.
Boundary markers: None identified in the provided instructions to separate data from commands.
Capability inventory: Extensive shell execution capabilities including system service control, firewall modification, and file system management across multiple scripts.
Sanitization: There is no evidence of sanitization or filtering applied to the ingested log or error data.

troubleshooting