The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the obsidian vault info=path command to programmatically determine the file system location of the user's knowledge base.
[DATA_EXFILTRATION]: During the research phase, the agent extracts keywords and phrases from the user's private vault notes and uses them to perform external web searches. This creates a data flow where potentially sensitive topical interests or internal project names are transmitted to search engines and LLM providers.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes untrusted data from the vault.
Ingestion points: Files are read from Clippings/, Daily Notes/, Projects/, Coding/, Meetings/, and Topics/ folders as specified in SKILL.md (Phases 0 and 1). The Clippings/ folder is particularly high-risk as it typically contains raw content saved from external websites.
Boundary markers: The instructions do not define delimiters or protective framing to isolate vault content from the agent's internal logic.
Capability inventory: The skill possesses file system read/write access, the ability to execute shell commands via the Obsidian CLI, and network access for web research (Phase 3).
Sanitization: No validation or escaping mechanisms are applied to the ingested content before it is processed by the model to determine future actions.

topic-pulse