The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze data from F5 system logs via the show_logs_tool and configuration details via the list_tool. Since logs can contain content derived from external network traffic (such as HTTP headers or URIs), they represent a vector for indirect prompt injection.
Ingestion points: System logs accessed via show_logs_tool and iRule/object configurations via list_tool.
Boundary markers: The skill does not provide explicit delimiters or instructions to the agent to disregard potential commands found within the log data.
Capability inventory: The agent has significant capabilities to modify system state, including update_tool, create_tool, and delete_tool.
Sanitization: There is no evidence of sanitization or filtering of log data before it is presented to the agent for analysis.
[COMMAND_EXECUTION]: The skill instructions rely on executing shell commands that incorporate environment variables for critical components.
Evidence: The skill uses the pattern python3 $MCP_CALL "python3 -u $F5_MCP_SCRIPT" to invoke its tools. This execution flow depends on the safety of the scripts pointed to by $MCP_CALL and $F5_MCP_SCRIPT.
[DYNAMIC_EXECUTION]: The skill facilitates the modification of iRules, which are Tcl scripts executed by the F5 BIG-IP system.
Evidence: The update_tool documentation provides examples of updating the apiAnonymous field with Tcl code blocks (e.g., the uri_routing example). The agent generates this logic and applies it to the infrastructure at runtime.

f5-troubleshoot