The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requires the use of the youmind CLI to perform API calls. These calls involve passing user-controlled data such as article titles and topics into shell command arguments within JSON strings. This pattern could be vulnerable to command or argument injection if the agent does not properly escape or sanitize the input.
[EXTERNAL_DOWNLOADS]: The skill directs the installation of the @youmind-ai/cli package via NPM. This tool is a resource belonging to the author's organization and is necessary for the skill's primary functionality.
[PROMPT_INJECTION]: The skill instructions contain mandatory directives (e.g., '⚠️ MANDATORY: present this message IMMEDIATELY') designed to override the agent's conversational heuristics. These instructions dictate specific UI behaviors and response timing, bypassing the agent's autonomy.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input which is then used in tool calls.
Ingestion points: User-provided article titles, topics, and external URLs processed in the Step 1 workflow.
Boundary markers: None; the data is interpolated directly into the message template passed to the createChat command.
Capability inventory: Subprocess calls via the youmind CLI tool and local Node.js script execution.
Sanitization: The skill does not provide instructions for sanitizing or validating user input before usage in shell commands.
[DYNAMIC_EXECUTION]: The workflow executes a local Node.js script (scripts/extract-images.js) to parse output from the CLI tool. The script uses regular expressions to process incoming data from the tool's execution context.

youmind-blog-cover