procurement-offer-review

Pass

Audited by Gen Agent Trust Hub on Aug 14, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill implements strong security best practices, such as maintaining vendor files as read-only, using versioned copies for internal documents, and enforcing factual verification against source documents.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external content in the form of supplier offers (Word and PDF).
  • Ingestion points: Supplier technical and financial offers are read during the OFFER-REVIEW mode.
  • Boundary markers: The skill instructions enforce a strict output schema ("Evidence Cards") and a "Finance-to-confirm" convention, which minimizes the risk of the agent following instructions embedded within vendor documents.
  • Capability inventory: The skill utilizes local file system access for reading and writing document files (.docx, .xlsx, .pdf) using the scripts/add_offer_comments.py script and methods described in references/docx-xlsx-mechanics.md.
  • Sanitization: Content is processed via standard libraries (python-docx, openpyxl, pypdf) which focus on structural extraction rather than executable content.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 14, 2026, 08:34 PM
Security Audit — agent-trust-hub — procurement-offer-review