Automated penetration testing for AI agents. Test against 67+ threat patterns including prompt injection, jailbreaks, and data exfiltration—before attackers do.
from agent_trust import RedTeam
redteam = RedTeam(api_key="ta_xxx...")
# Run security scan against your agent
result = redteam.scan(
"https://my-agent.com/chat",
mode="comprehensive"
)
print(f"Security Score: {result.security_score}/100")
print(f"Risk Level: {result.risk_level}")
if result.has_critical_issues:
print("⚠️ Critical vulnerabilities found!")
for vuln in result.vulnerabilities:
print(f" - [{vuln.severity}] {vuln.threat_name}")
# Export HTML report
redteam.export(result, "security-report.html")Red Team scans your agent against 67+ attack patterns. Find vulnerabilities before attackers do.
Guard protects your agent at runtime. Scan content before your agent processes it.
Analytics dashboard shows threats blocked, patterns detected, and security trends.
Red Team tests against all major AI agent attack vectors
Block insecure agents from reaching production
name: AI Agent Security Scan
on: [pull_request]
jobs:
security:
runs-on: ubuntu-latest
steps:
- name: Run Red Team Scan
uses: trustagents/redteam-action@v1
with:
target_url: ${{ secrets.AGENT_URL }}
api_key: ${{ secrets.TRUSTAGENTS_API_KEY }}
fail_threshold: high
output_file: security-report.html