The Agent Skills Directory

[SAFE]: The skill consists entirely of markdown instructions and reference material. It does not include scripts, binaries, or commands that interact with the network or sensitive system files.
[PROMPT_INJECTION]: The skill facilitates the processing of user-supplied data from Obsidian notes (frontmatter, tags, and file properties). While legitimate, this creates a surface for indirect prompt injection where malicious instructions embedded in a note could attempt to override agent behavior during file processing.
Ingestion points: Data is read from note frontmatter (note.author), file metadata (file.properties), and folder structures (file.inFolder).
Boundary markers: No delimiters or instructions are used to distinguish untrusted note content from agent instructions.
Capability inventory: The agent's primary capability within this skill is the creation and modification of workspace .base files.
Sanitization: The reference material provides an escapeHTML() function, though its application to ingested data is not strictly enforced in the workflow.

obsidian-bases