pdf-reader

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • Indirect Prompt Injection (LOW): The skill extracts text from external PDFs and feeds it to the agent without providing clear boundaries or sanitization.
  • Ingestion points: Data is ingested via the convert_to_markdown tool (from URIs) and the extract_text Python utility (from local paths) as described in SKILL.md.
  • Boundary markers: Absent. The instructions do not define delimiters or provide warnings to the agent to ignore instructions found within the PDF text.
  • Capability inventory: The agent can write to notes (save_to_notes) and manipulate local files, providing a potential path for injected instructions to affect the user's system.
  • Sanitization: Absent. No filtering or sanitization of the extracted PDF text is performed before it is processed by the AI.
  • External Downloads (LOW): The convert_to_markdown tool (attributed to microsoft-mar) accepts http:// and https:// URIs to download and process remote PDF files. While from a trusted source, downloading arbitrary content poses a minor risk of data exposure or processing unexpected inputs.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:41 PM