dingdawg-shield

AI Security Scanning -- vulnerability detection, governance checks, and trust scoring for AI systems.

Quick Start

# Claude Code
claude mcp add dingdawg-shield npx dingdawg-shield

# Or run directly
npx dingdawg-shield

What You Get

Free: Security Scan + Governance Check (no account needed)

• AI vulnerability scanning: prompt injection, data leakage, unsafe deserialization
• Governance assessment across 5 pillars: transparency, accountability, fairness, safety, privacy
• Risk scoring with actionable remediation steps
• Results in under 60 seconds

Paid: Trust Score ($0.05/check)

• Quantified 0-100 trust score for any AI system
• Breakdown across reliability, safety, transparency, fairness, accountability
• Percentile ranking against industry benchmarks
• Trust badge for verified systems

Tools

| Tool | Description | Cost | |------|-------------|------| | security_scan | Scan code for AI security vulnerabilities | Free (1/day quick), Paid (standard/deep) | | governance_check | Check AI system governance across 5 pillars | Free | | trust_score | Quantified trust score for an AI agent/system | $0.05/check |

Scan Depths

| Depth | Coverage | Cost | |-------|----------|------| | Quick | Prompt injection, data leakage, unsafe deserialization, missing validation | Free (1/day) | | Standard | + dependency audit, configuration review, access control analysis | API key required | | Deep | + full attack surface mapping, remediation playbook, compliance mapping | API key required |

Setup

1. Free tier: No setup needed. Install and run security_scan or governance_check.
2. Paid features: Get your API key at dingdawg.com/developers
3. Set environment variable: export DINGDAWG_API_KEY=your_key

Environment Variables

| Variable | Required | Description | |----------|----------|-------------| | DINGDAWG_API_KEY | For paid tools | API key from dingdawg.com/developers | | DINGDAWG_API_URL | No | Custom API endpoint (default: https://api.dingdawg.com/v1) |

Why DingDawg Shield?

Every scan and check is governed -- meaning every assessment is capability-checked, policy-evaluated, and receipted. You get a governance proof with every result proving it was generated under auditable controls.

AI security is not optional. Prompt injection, data leakage, and model extraction are real attack vectors. Shield catches them before your users do.

License

MIT