skills/build-plus/skills/bid-leveling/Gen Agent Trust Hub

bid-leveling

Pass

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: LOWEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [External Downloads] (LOW): The script scripts/generate_pdf.py utilizes the reportlab library for PDF generation. This is a well-established and trusted package within the Python ecosystem, posing negligible risk in standard environments.
  • [Indirect Prompt Injection] (LOW): The skill is designed to process external, untrusted documents (bids, proposals). There is an inherent risk that an attacker could embed instructions in a bid document to manipulate the AI's extraction logic (e.g., hiding exclusions or inflating values).
  • Ingestion points: scripts/generate_pdf.py processes the structured output derived from bid documents.
  • Boundary markers: No explicit delimiters are defined in the instructions to prevent the model from following instructions found within the bid text.
  • Capability inventory: The skill can write files to the local system (the generated PDF report) but lacks network or shell execution capabilities.
  • Sanitization: The PDF generator uses reportlab.platypus.Paragraph which renders text but does not execute it; however, there is no logic to sanitize the content extracted from the bids for deceptive language.
Audit Metadata
Risk Level
LOW
Analyzed
Feb 16, 2026, 01:05 PM