The Agent Skills Directory

[DATA_EXFILTRATION]: The skill instructions direct the agent to read server.cfg and follow exec lines to other configuration files. The skill explicitly acknowledges that these files contain sensitive secrets such as mysql_connection_string, sv_licenseKey, and steam_webApiKey. Reading these files into the agent's context constitutes a data exposure risk, as the content could be leaked if the agent is compromised or via a subsequent interaction.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data by scanning the resources/ directory and reading fxmanifest.lua files for every installed resource.
Ingestion points: fxmanifest.lua files from third-party resources and the server.cfg file.
Boundary markers: The skill does not define clear delimiters or instruct the agent to disregard potential instructions embedded within these files.
Capability inventory: The agent has the ability to read arbitrary files within the server directory and write to configuration files in the user's home directory (~/.config/ox/).
Sanitization: There is no evidence of sanitization or strict validation of the fields read from external manifests (e.g., version, locale) before they are used to generate output.

setup-ox