BlackBox Labs — VS Code extension to detect provenance of AI Code!

The BlackBox Labs IDE Extension was created as part of a capstone group project for DECO3801 (Design Computing Studio) at The University of Queensland in Semester 2, 2025. As the semester has closed, our servers no longer support running the extension. We thank our early users for their time to support us!

Flag code for AI-provenance.

What it does

• Analyze file or workspace
• Analyze a commit (added lines only) from a list or an interactive commit tree.
• Code highlighting + Markdown report covering AI-Detection.
• Custom thresholds via bbx-policy.yml.

How to run it

• Shortcut: Ctrl+Alt+B opens the action menu (if unbound).

• GUI:

  • Status Bar (bottom-left): “BlackBox Policy Check” button.
  • Editor title bar (top-right): flask icon.

• Pick: Audit file, Audit whole repo, Pick commit from list, or Pick from tree.

Requirements

• Open folder/workspace; Git for commit features.
• Internet access
• Languages: Python, Java.

Policy file example

# BlackBox Labs Default Policy
rules:
  # --- AI Provenance Thresholds ---
  - id: ai_low_risk
    type: ai-threshold
    severity: low
    threshold: 0.30
    description: Minor stylistic or token-level AI influence. Logged for provenance only

  - id: ai_medium_risk
    type: ai-threshold
    severity: medium
    threshold: 0.55
    description: Mixed authorship. Light human review recommended

  - id: ai_high_risk
    type: ai-threshold
    severity: high
    threshold: 0.75
    description: Strong AI signature. Requires manual inspection

  - id: ai_critical_risk
    type: ai-threshold
    severity: critical
    threshold: 0.90
    description: Near-pure model generation. Requires audit for logic consistency, licensing, and attribution