geo-report-pdf

Pass

Audited by Gen Agent Trust Hub on Aug 9, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill installs the reportlab library from the Python Package Index (PyPI), which is a standard and well-known dependency for PDF generation.
  • [COMMAND_EXECUTION]: The skill executes a local Python script located at ~/.claude/skills/geo/scripts/generate_pdf_report.py to process audit data and generate a PDF file.
  • [PROMPT_INJECTION]: The skill processes data from existing markdown audit reports (e.g., GEO-*.md) in the local environment, creating a surface for indirect prompt injection.
  • Ingestion points: The agent parses markdown files such as GEO-CLIENT-REPORT.md or GEO-AUDIT-REPORT.md from the current working directory to extract scoring data and findings.
  • Boundary markers: The skill instructions do not provide explicit delimiters or instructions to the agent to treat the parsed content strictly as data, potentially allowing embedded instructions to be obeyed.
  • Capability inventory: The skill involves writing to the file system (/tmp/geo-audit-data.json) and executing shell commands through the PDF generation script.
  • Sanitization: No sanitization or schema validation is applied to the data parsed from the markdown files before it is processed by the generation script.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 9, 2026, 03:25 AM
Security Audit — agent-trust-hub — geo-report-pdf