template-engine

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data files such as user-provided CSV spreadsheets and document templates to automate generation workflows. This creates an indirect prompt injection attack surface if the text contents of those files contain malicious instructions that the agent might interpret as commands.\n
  • Ingestion points: Reads external files like data/contacts.csv and template documents through the provided Python code samples in SKILL.md.\n
  • Boundary markers: Absent. No data separation or explicit safety instructions are outlined to guard against executable instructions within data values.\n
  • Capability inventory: Restricted to document manipulation and file system writes via docxtpl, openpyxl, and csv. No network requests or arbitrary subprocess executions are defined.\n
  • Sanitization: Absent. The template rendering logic assumes data values are safe and does not strip or neutralize prospective language-model injection strings.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 08:53 PM
Security Audit — agent-trust-hub — template-engine