tech-debt-analyzer

Pass

Audited by Gen Agent Trust Hub on Feb 23, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • Indirect Prompt Injection (LOW): The dependency analysis script processes external package.json files and prints their contents into a report. If these files contain malicious instructions in metadata fields like package names or descriptions, they could influence the agent's subsequent behavior.\n
  • Ingestion points: scripts/analyze_dependencies.py (line 34) reads target JSON files provided by the user or agent.\n
  • Boundary markers: Absent; the script outputs raw package names and versions into the Markdown report without delimiters or safety warnings for the agent.\n
  • Capability inventory: The script is limited to local file reads and JSON parsing; no subprocess execution, dynamic evaluation, or network operations were identified.\n
  • Sanitization: Absent; input strings are parsed as JSON and printed directly to the report without escaping or validation.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 23, 2026, 05:29 AM