soia-pkm-clip-web

Pass

Audited by Gen Agent Trust Hub on Sep 1, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill extracts data from arbitrary web URLs using readability libraries, exposing a surface for instructions embedded in external pages. Ingestion point: Web content extraction mentioned in SKILL.md. Boundary markers: The skill mandates a manual quality review of captured content before vault storage to mitigate extraction errors. Capability inventory: File system writing to Obsidian and execution of system commands for PDF processing. Sanitization: Relies on manual human review rather than automated filtering of malicious instructions.
  • [COMMAND_EXECUTION]: The skill utilizes system-level commands to automate features. Specifically, it uses osascript (macOS) to control the Obsidian UI for PDF generation and calls CLI utilities pdfinfo, pdftoppm, and pdftotext to verify output integrity.
  • [EXTERNAL_DOWNLOADS]: The skill depends on external Python packages trafilatura and readability-lxml for its core extraction functionality. It also references installation procedures involving npx and plugin marketplace commands from vendor-managed repositories.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 1, 2026, 03:51 AM
Security Audit — agent-trust-hub — soia-pkm-clip-web