The Agent Skills Directory

[COMMAND_EXECUTION]: The Tools/VaultBuilder.py script uses the subprocess.run and os.startfile functions to open the Obsidian application via URI schemes and check for running processes. These operations are cross-platform and are used strictly for vault management and verification.
[COMMAND_EXECUTION]: The skill executes shell commands in Workflows/Setup.md to create the vault's folder structure, establish symlinks for companion skills, and open the vault directory in Obsidian.
[SAFE]: The skill modifies the ~/.claude/CLAUDE.md file through the inject-global command in Tools/VaultBuilder.py. This is a persistence mechanism used to load the vault's context into the agent globally, which is a standard configuration for this platform and is explicitly presented to the user as an optional choice.
[SAFE]: The scripts/process_docs_to_obsidian.py script facilitates the ingestion of external documents into the vault. While this creates a surface for indirect prompt injection if the imported files contain malicious instructions, the script itself only performs file copying and metadata addition without executing or interpreting the file content.

vault-setup