The Agent Skills Directory

[SAFE]: The skill documentation and referenced files contain legitimate syntax guides for Obsidian Markdown. No malicious code, obfuscation, or exfiltration patterns were detected.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied markdown and external web data (Ingestion points: SKILL.md Protocol 4, user notes). While explicit boundary markers are not provided in the syntax documentation, the skill mandates redacting secrets/PII and treating inputs as hostile (Sanitization: SKILL.md Protocol 4). The agent context includes file operations and project-scoped tool execution (Capabilities: SKILL.md Protocols 1 & 3). This surface is inherent to the skill's purpose and is mitigated by the specified hygiene protocols.
[SAFE]: The 'GLOBAL PROTOCOLS' section defines operational constraints that enhance the safety and predictability of the agent, including the use of deterministic execution loops and the principle of least privilege.

obsidian-markdown