The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill installs the @youmind-ai/cli package from the NPM registry. This is a vendor-owned package from YouMind-OpenLab and is a core requirement for the skill's functionality.
[COMMAND_EXECUTION]: The skill makes extensive use of the youmind CLI to create chats, poll for status, and retrieve messages. It also uses node with the -e flag to process JSON output from these commands.
[CREDENTIALS_UNSAFE]: The skill requires a YOUMIND_API_KEY environment variable. The documentation includes best-practice advice, instructing the user to set the key in their configuration rather than pasting it directly into the chat.
[DATA_EXFILTRATION]: User-provided presentation topics and outlines are transmitted to youmind.com for processing. This is expected behavior as it is the service's primary function, and the data is sent to the official vendor domain.
[PROMPT_INJECTION]: The skill includes instructions for the agent to surface a mandatory onboarding message and provide progress updates. These are functional instructions and do not attempt to bypass core agent safety guidelines.
[PROMPT_INJECTION]: (Category 8
Indirect Prompt Injection Surface)
Ingestion points: User-provided presentation topics, outlines, and key points are used as inputs for shell commands in SKILL.md (Step 3).
Boundary markers: The instructions do not define boundary markers or delimiters (e.g., """ or ===) to separate user-provided content from the JSON payload in the shell command.
Capability inventory: The skill has the capability to execute shell commands via youmind, npm, and node, and it can perform network operations via the CLI.
Sanitization: There is no explicit sanitization or escaping logic described for the user-provided topic before it is interpolated into the single-quoted shell command youmind call createChat '{"message":"<topic/outline>"}'. This creates a surface where single quotes in a user's topic could break the JSON structure or shell command execution.

youmind-slides-generator