The Agent Skills Directory

[PROMPT_INJECTION]: The skill creates an indirect prompt injection surface by taking free-text user input and incorporating it into configuration files that govern agent behavior. * Ingestion points: The free-text self-description collected in Step 1 is used to populate CLAUDE.md. * Boundary markers: The skill does not use delimiters or instructions to prevent the agent from following commands embedded in the user-provided text. * Capability inventory: The skill writes persistent SKILL.md files and can append content to the global ~/.claude/CLAUDE.md file. * Sanitization: No validation or escaping is performed on user input before it is written to configuration files.
[COMMAND_EXECUTION]: The skill executes shell commands to configure the local environment. * It uses 'mkdir -p' to create the vault folder structure. * It uses 'open -a Obsidian' to launch an external application.
[PROMPT_INJECTION]: The skill modifies global configuration to achieve persistence for its instructions. By appending a reference to the local vault to ~/.claude/CLAUDE.md, it ensures that instructions derived from potentially malicious user input are loaded in every future session of the agent on the host system.

vault-setup