obsidian-brain

Pass

Audited by Gen Agent Trust Hub on Aug 3, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill follows security best practices by including explicit instructions to avoid storing passwords, API keys, and other sensitive credentials within the Obsidian vault notes.- [SAFE]: Technical analysis of the provided Python scripts (scripts/link_graph.py and scripts/person_matches.py) confirms they are limited to local filesystem operations using standard Python libraries. The scripts do not contain any network communication code, data exfiltration patterns, or dangerous dynamic execution functions.- [SAFE]: The skill identifies an indirect prompt injection surface through its primary function of ingesting untrusted external data (PDFs, transcripts, and exports). \n
  • Ingestion points: Data entering through references/import-workflows.md into the sources/ directory. \n
  • Boundary markers: Absent; there are no specific prompt delimiters to isolate imported content. \n
  • Capability inventory: The agent possesses filesystem access to the vault; scripts are restricted to read-only graph analysis. \n
  • Sanitization: The skill implements sanitization by encouraging the conversion of raw documents into structured summaries and atomic notes, which requires agent reasoning and reduces the likelihood of direct instruction following from the source data.- [SAFE]: All external resource references, including GitHub links and package installation commands, point to the author's verified vendor repositories (Thibaultbm), representing standard and expected resource access.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 3, 2026, 08:40 PM