The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill contains a hardcoded credential placeholder (admin:password) within a sample command for the Asterisk Manager Interface (AMI). While intended as an example, it represents a risk for accidental use or exposure.- [COMMAND_EXECUTION]: The instructions direct the agent to utilize docker exec for executing shell commands, interacting with the Asterisk CLI, and running AGI scripts stored on the filesystem. This grants the agent administrative-level interaction with the container environment.- [DATA_EXFILTRATION]: The skill is designed to read from the primary system configuration database (/cf/conf/mikopbx.db) and follow the main system logs (/var/log/asterisk/full), which may contain sensitive call routing and metadata.- [PROMPT_INJECTION]: The skill monitors log files that can contain external, unvalidated inputs such as Caller ID strings or SIP header content, creating a surface for indirect prompt injection.
Ingestion points: /var/log/asterisk/full (monitored via tail -f).
Boundary markers: None; the skill does not instruct the agent to distinguish between system logs and potential embedded instructions.
Capability inventory: docker exec (shell/CLI), sqlite3 (database read), and curl (localhost interaction).
Sanitization: None; the agent processes raw log output without filtering.

asterisk-tester