The Agent Skills Directory

[DATA_EXFILTRATION]: The skill is configured to read from and write to specific local filesystem paths, including the user's home directory (/Users/joel/Vault/Resources/videos/). While this is the primary purpose of the skill for the author, accessing the home directory is a sensitive operation that could lead to unintended data exposure if the agent is misdirected.
[COMMAND_EXECUTION]: The skill's documentation explicitly includes shell commands (ls) as part of the reference for interacting with the content system. Although these are intended as examples, they demonstrate the agent's capability to interact with the system shell.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it processes untrusted content from external markdown files.
Ingestion points: The agent reads content from arbitrary markdown files located in the /Users/joel/Vault/Resources/videos/ directory.
Boundary markers: There are no explicit boundary markers or instructions (such as 'treat the following as data only') to prevent the agent from obeying instructions hidden within the source notes.
Capability inventory: The agent has the ability to read local files, transform content using its internal writing style logic, and write new files to the apps/web/content/ directory.
Sanitization: No sanitization logic is present to filter or escape instructions embedded within the source markdown files before they are processed by the LLM.

video-note