The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Grep tool to search for specific frontmatter fields (title, tags, aliases, summary) and section content within Markdown files in the configured vault path. This is used to build candidate sets and extract relevant context without reading entire files.\n- [PROMPT_INJECTION]: As a retrieval-based skill, it is subject to indirect prompt injection from the content it processes. \n
Ingestion points: Reads index.md, hot.md, and arbitrary page bodies from the Obsidian vault directory.\n
Boundary markers: None identified; wiki content is synthesized directly into answers.\n
Capability inventory: Performs file searches with Grep, reads file content with Read, and appends query logs to log.md.\n
Sanitization: None identified; the agent relies on its own synthesis logic to interpret retrieved data.\n- [SAFE]: The skill accesses local configuration files including .env and ~/.obsidian-wiki/config to determine the OBSIDIAN_VAULT_PATH. This is standard initialization for a local knowledge-base tool and does not involve the exfiltration of credentials or sensitive data to external sources.

wiki-query