The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from vault pages to generate its readouts, creating a surface where embedded instructions could influence the agent.
Ingestion points: Reads from arbitrary vault pages, index.md, and hot.md during the retrieval phase.
Boundary markers: The skill instructs the agent to honor visibility tags (e.g., visibility/internal) and filtered-mode phrases, but does not implement strict delimiters or 'ignore' instructions for the main content of retrieved pages.
Capability inventory: The skill can read any file in the vault (excluding specific metadata folders), write new readouts to _readouts/, append to log.md, and modify hot.md. It also executes search commands.
Sanitization: The skill uses deterministic, filesystem-safe slugs for generated files and requires specific voice skeletons, which limits the influence of external data on the overall structure.
[COMMAND_EXECUTION]: The skill utilizes rg (ripgrep) and qmd (Query Markdown) command-line tools to search the vault. These are functional requirements for the skill's retrieval mechanism.

wiki-narrate