ai-deep-research

Pass

Audited by Gen Agent Trust Hub on Aug 12, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides a Python script scripts/citation_verifier.py intended for local citation validation. A technical review of the script confirms it uses only standard library modules and performs text analysis on JSONL files without spawning subprocesses or executing arbitrary commands.
  • [EXTERNAL_DOWNLOADS]: The skill references multiple external URLs for documentation and research tools, such as official developer portals for OpenAI, Google, Anthropic, and Perplexity, as well as several reputable GitHub repositories. All identified external links point to well-known technology services or trusted developer resources, consistent with the skill's instructional purpose.
  • [PROMPT_INJECTION]: The skill's primary function involves retrieving and processing external web content, creating a surface for indirect prompt injection. The skill mitigates this by mandating a 'Verifier' subagent pattern that operates in an isolated context and employing 'Hostile-source detection' techniques to filter adversarial SEO or AI-generated content before synthesis.
  • [DATA_EXFILTRATION]: No evidence of hardcoded credentials, sensitive file access, or unauthorized network exfiltration was found. The skill follows best practices by recommending local file storage for session persistence and plan management.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 12, 2026, 09:09 PM
Security Audit — agent-trust-hub — ai-deep-research