The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the bash tool to execute curl commands. Assembling shell commands that incorporate user-provided content (such as messages or questions) is a significant security risk if not handled by a secure tool-calling layer that performs strict escaping.
[EXTERNAL_DOWNLOADS]: The skill communicates with the LinkAI API at api.link-ai.tech. This is the official domain for the service and is required for the skill to function.
[PROMPT_INJECTION]: The skill architecture includes an indirect prompt injection surface where it dynamically updates its own description based on application names and descriptions provided in config.json.
Ingestion points: The apps list within config.json.
Boundary markers: None identified; the configuration data is directly incorporated into the agent's skill description.
Capability inventory: Uses the bash tool for network operations and API calls.
Sanitization: No evidence of sanitization or validation for the content of the application descriptions in the configuration file.

linkai-agent