sophnet-pdf

Pass

Audited by Gen Agent Trust Hub on Mar 6, 2026

Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [DATA_EXFILTRATION]: The script scripts/upload_file.sh facilitates the upload of generated PDF files to a remote storage service using the sophnet-tools vendor package. While this is the intended delivery mechanism, it represents a network-based data transfer capability.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its core functionality of reading and extracting text from untrusted PDF files.
  • Ingestion points: Untrusted data is ingested through scripts/extract_pdf_content.py, scripts/extract_form_structure.py, and scripts/extract_form_field_info.py which parse user-provided PDF documents.
  • Boundary markers: The skill does not implement delimiters or 'ignore' instructions when presenting extracted text from PDFs to the agent.
  • Capability inventory: Across its scripts, the skill can perform network operations (via scripts/upload_file.sh), execute subprocesses (via uv run), and write files to the local filesystem.
  • Sanitization: There is no evidence of sanitization or filtering of extracted PDF content before it is processed by the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 6, 2026, 03:34 AM