🛡️ AgentOps

AI agent security scanner — detect secrets before they reach cloud models.

Scans your codebase, config files, and AI agent workspaces for exposed API keys, passwords, tokens, and other credentials that could be sent to LLM providers.

Install

npm install -g agentops
# or
npx agentops scan .

Usage

# Scan current directory
agentops scan .

# Scan specific path
agentops scan ~/.openclaw/workspace

# Output as JSON (for piping/API)
agentops scan . --format json

# Only show high/critical findings
agentops scan . --severity high

# Disable entropy detection (faster)
agentops scan . --no-entropy

# Add custom exclude patterns
agentops scan . --exclude temp --exclude backup

What It Detects

• AWS/GCP/Azure cloud credentials
• GitHub/Stripe/SendGrid API tokens
• Database connection strings
• Private keys (SSH, RSA, PEM)
• JWT tokens and bearer auth
• AI provider keys (OpenAI, Anthropic, Ollama, HuggingFace, etc.)
• OpenClaw gateway credentials and plugin keys
• Oracle Cloud OCIDs and API keys
• High-entropy strings that look like secrets

Exit Codes

• 0 — No secrets found
• 1 — Secrets detected (critical or high)
• 2 — Fatal error

Examples

🛡️  AgentOps Security Scanner v0.1.0

🔴 CRITICAL (2)
  /project/.env:3 — AWS Access Key ID
    AKIA...X7E2
    ID: AWS_ACCESS_KEY_ID
  /project/config.json:47 — OpenAI API Key
    sk-proj...9Kp2
    ID: OPENAI_API_KEY

🟡 HIGH (1)
  /project/.env:7 — Stripe Secret Key
    sk_liv_...8f3G
    ID: STRIPE_SECRET_KEY

Scan Summary
🔴 Critical: 2  🟡 High: 1  🟠 Medium: 0  🔵 Low: 0
Files scanned: 47

License

MIT