nusantara-corpus-pdf-extractor

Pass

Audited by Gen Agent Trust Hub on Sep 2, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill extracts text from user-provided dictionary PDFs and incorporates this potentially untrusted data into prompts and tool inputs, such as web search queries. A malicious document could attempt to influence agent behavior through embedded instructions. Ingestion points: pdf_parser.py extracts text from digital and OCR layers of PDFs. Boundary markers: Absent; the agent is not instructed to use specific delimiters or ignore instructions within the extracted data. Capability inventory: The agent executes Python scripts (cli.py) with file system access and can perform web_search based on extracted headwords. Sanitization: No specific filtering or sanitization is performed on extracted strings before they are interpolated into tool queries or context.
  • [EXTERNAL_DOWNLOADS]: The README directs users to install the vendor-owned nusantara-corpus-extractor package via pip and the skill itself via npx skills add. It also recommends installing the well-known tesseract OCR engine via Homebrew.
  • [COMMAND_EXECUTION]: The skill requires the agent to run local Python scripts and CLI tools provided in the repository to orchestrate the extraction pipeline.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 2, 2026, 01:58 PM