travel-dossier

Pass

Audited by Gen Agent Trust Hub on Jul 24, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses local command-line tools to manage the PDF creation and quality assurance process.
  • The script scripts/build.sh executes a headless browser (Chrome, Chromium, or Microsoft Edge) to convert HTML content into a PDF document.
  • The script scripts/check_overflow.py invokes the pdftoppm utility to render PDF pages into images for automated visual inspection of page footers.
  • [EXTERNAL_DOWNLOADS]: The skill performs external network requests to retrieve content required for the dossier.
  • The scripts/fetch_images.py script downloads image files from Wikimedia Commons using its public API.
  • The instructions in SKILL.md guide the agent to use curl with the -k flag to bypass certificate validation for specific government domains (gob.pe) known to have incomplete certificate chains.
  • [PROMPT_INJECTION]: The skill processes external PDF files, creating a potential surface for indirect prompt injection.
  • Ingestion points: Text is extracted from user-provided travel agency itineraries and ticket confirmation PDFs in Step 2.
  • Boundary markers: There are no explicit markers or instructions to isolate extracted text from the agent's control logic.
  • Capability inventory: The skill can execute local binaries (Chrome, pdftoppm), write files to the system, and make network requests (Wikimedia API).
  • Sanitization: There is no evidence of filtering or sanitizing the content extracted from PDFs before it is used in the dossier generation pipeline.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 24, 2026, 12:25 PM
Security Audit — agent-trust-hub — travel-dossier