All work

• CloudSheriff: a read-only CIS drift monitor for AWS2026

  A read-only monitor that finds CIS drift in AWS, traces the change in CloudTrail, and uses an LLM to explain alerts, not decide them.

  • AWS
  • CIS
  • Drift detection
  • LLM guardrails
  • OpenTofu
• Hermes: a multi-agent security review assistant in Slack2026

  One Slack entry point and one orchestrator route security reviews to specialist reviewer skills and return an evidence-based report.

  • Multi-agent
  • Slack
  • Security review
  • Compliance
  • Hermes Agent
• Agent guardrails: making AI coding agents safe to use2026

  How we make AI coding agents safe to use: verification gates, test-integrity hooks, an allowlisted MCP server and a post-mortem.

  • AI agents
  • Guardrails
  • DevSecOps
  • MCP
  • Incident response
• Automating firewall policy change review2025

  A browser automation that walks a firewall policy review queue, opens each revision and saves screenshot evidence with a step log.

  • automation
  • firewall
  • change-review
  • Selenium
  • Python
• Finding security-tool coverage gaps by reconciling the CMDB2025

  Reconcile the CMDB against security-tool exports and show the coverage gaps as a Sankey diagram. The public demo uses synthetic data.

  • CMDB
  • asset-inventory
  • coverage
  • Sankey
  • Python
• A repeatable security audit for your own Mac2026

  A repeatable, evidence-based Mac audit: persistence items, privacy grants, browser extensions, and a second round to verify fixes.

  • macOS
  • hardening
  • persistence
  • privacy
  • KnockKnock