obsidian

Pass

Audited by Gen Agent Trust Hub on Sep 6, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill is designed for local file system interaction within the scope of user-defined Obsidian vaults. It does not contain network-facing tools, hardcoded secrets, or mechanisms for privilege escalation.
  • [INDIRECT_PROMPT_INJECTION]: The skill provides an attack surface for indirect prompt injection as it processes content from external markdown files. 1. Ingestion points: Notes are ingested through obsidian_read and searched via obsidian_search. 2. Boundary markers: The skill includes instructions to treat content as user writing and to cite sources, though no physical delimiters are enforced. 3. Capability inventory: The toolset is limited to local file operations (search, read, create, append). 4. Sanitization: No explicit sanitization of note content is present. As this is an inherent risk of document-processing tools and the skill follows best practices for user intent, the severity is assessed as safe.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 6, 2026, 09:31 AM
Security Audit — agent-trust-hub — obsidian